technical report on the dvb-h and dvb-sh
TRANSCRIPT
1� HSC
Technical Report on the DVB-H and DVB-SH standards
v 0.2
HSC Restricted
D -8, Infocity – II ,Sector 33,
Gurgaon,Haryana.
India
2� HSC
PROPRIETARY NOTICE
All rights reserved. This publication and its contents are proprietary to Hughes Systique Corporation. No part of this publication may be reproduced in any form or by any means without the written permission of Hughes
Systique Corporation, 15245 Shady Grove Road, Suite 330, Rockville, MD
20850.
Copyright © 2006 Hughes Systique Corporation
3� HSC
Revision History
Rev. ID
Date of issue
Author
Approver Change Page No.
Action Taken (Addition, modification, deletion)
Rationale for change
1 20/6/2007
Partho Choudhury
Initial draft
2 25/7/2007
Partho Choudhury
Additions as per
suggestions by Channasandra Ravishankar
3
4� HSC
TABLE OF CONTENTS
SECTION PAGE
1.0 TECHNICAL OVERVIEW .............................................................................................................6 1.1 HISTORICAL BACKGROUND .....................................................................................................6 1.2 MOTIVATION ................................................................................................................................6
2.0 SYSTEM ARCHITECTURE ..........................................................................................................7 2.1 DVB-H ...........................................................................................................................................7 2.2 DVB-SH ........................................................................................................................................12 2.3 OFDM IN SATELLITE LINKS ......................................................................................................13 2.3.1 Line of Sight (LOS)...................................................................................................................13 2.3.2 Amplifier non-linearities and Output Back-Off (OBO) ..........................................................13 2.3.3 Suitable remedies possible .....................................................................................................13 2.4 SFN NETWORKS VS. MFN NETWORKS ...................................................................................14 2.5 SYNCHRONIZATION OF SC AND CGC SIGNALS IN A SFN NETWORK ................................15 2.6 SYNCHRONIZATION IN DVB-SH-A ............................................................................................15 2.7 SYNCHRONIZATION IN DVB-SH-B ............................................................................................16 2.8 TR(A) AND TR(C) CLASS TRANSMITTERS ..............................................................................16 2.9 TR(B) CLASS TRANSMITTERS ..................................................................................................16
3.0 ARCHITECTURAL CONSTITUENTS ..........................................................................................19 3.1 CONSTITUENTS OF THE DVB-H SYSTEM ...............................................................................19 3.1.1 DVB-H and DVB-SH embedded within DVB-T .......................................................................19 3.1.2 The LLC/MAC layer ..................................................................................................................19 3.1.3 Encapsulation and packetization in the DLC layer ...............................................................21 3.1.4 The PHY layer ...........................................................................................................................21 3.1.5 Issues specific to DVB-SH ......................................................................................................23 3.1.6 Interleaving ...............................................................................................................................23 3.1.7 The Holy Grail of mobile reception ........................................................................................24 3.1.8 Modulation of the baseband signal ........................................................................................24 3.1.9 Advantages of Hierarchical Modulation – “Simulcast”........................................................25
4.0 SPECTRUM ALLOCATION AND RELATED BANDWIDTH ISSUES .........................................25 4.1 FREQUENCY BAND SELECTION FOR THE DVB-SH SYSTEM ...............................................27 4.2 DOPPLER CONSIDERATIONS ...................................................................................................28
5.0 MOBILE MULTICAST/BROADCAST SERVICE (MBMS) ...........................................................28 5.1 MULTICAST IP SERVICE ............................................................................................................28 5.2 BROADCAST IP SERVICE ..........................................................................................................29 5.3 BROADCAST IP SERVICE ..........................................................................................................30
6.0 DOWNLINK ANALYSIS – FRAMEWORK AND PARAMETERS................................................30 6.1 ASSUMPTIONS ............................................................................................................................30 6.2 SATELLITE COMPONENT ..........................................................................................................31 6.3 CARRIER FREQUENCY ..............................................................................................................32 6.4 COMPLEMENTARY GROUND COMPONENT ...........................................................................36
7.0 FADING CHARACTERISTICS OF DVB-H AND SC AND CGC FOR DVB-SH ..........................39 7.1 RICIAN FADING ...........................................................................................................................39 7.2 RAYLEIGH FADING .....................................................................................................................40 7.3 LOG NORMAL (SUZUKI) FADING ..............................................................................................40
5� HSC
8.0 PHY/MAC LAYER PERFORMANCE ...........................................................................................40 8.1 DOPPLER EFFECTS IN DVB-SH-A ............................................................................................41 8.2 VARIATION IN DOWNLINK TRIP DELAYS IN DVB-SH-B .........................................................41 8.3 CHOICE OF MODULATION SCHEME IN THE MAPPER ...........................................................41 8.4 PERFORMANCE CHARACTERISTICS OF MPE-FEC TIME INTERLEAVERS ........................42 8.5 PERFORMANCE CHARACTERISTICS OF NATIVE AND IN-DEPTH INTERLEAVERS ..........42 8.6 TRADEOFFS IN TIME-INTERLEAVER DESIGN IN THE MAC LAYER .....................................43
9.0 VIDEO AND AUDIO PROFILES SUPPORTED BY DVB-H AND DVB-SH .................................44
10.0 REFERENCES ............................................................................................................................45
6� HSC
1.0 TECHNICAL OVERVIEW The DVB-H and DVB-SH standards are the latest in a series of PHY and MAC layer specifications conceptualized and standardized by the Digital Video Broadcast (DVB) Forum. Earlier standards published by this body include the DVB-T, DVB-S and DVB-C standards, which were meant to broadcast digital video content over terrestrial, satellite and cable delivery systems, respectively. In fact, the latest DVB-H and DVB-SH are fully backward compatible with (and a superset of) the older DVB-T standard. 1.1 Historical Background The first attempt to introduce mobility in video broadcast systems was done under the aegis of the ACTS-MOTIVATE (MObile Television and InnoVATive rEceivers) project in 2000. This project showed that is technically feasible and financially viable to design and build a separate wireless broadcast system to deliver seamless “broadcast quality” TV signals to receivers which are “portable” (low speed and/or short range mobility). Two years later, the EU sponsored the Multimedia Car Platform (IST-MCP) project, which explored the behavior of antenna diversity reception, in addition to the frequency and time diversities already provided for by the DVB-T standard. In the 5 years since the MCP project, viewing habits of consumers have undergone a sea change, with the emphasis has shifted from “portable” devices (such as a laptop being lugged around the house, or driven around the block in a car) to “perpetually mobile” (medium to high speed and/or long range mobility) “low-power and low-cost” handheld devices. This has had a direct bearing on various design parameters of the DVB-H/DVB-SH system in comparison to the legacy DVB-T system. The outcome of this effort is the DVB-H standard, which was ratified in 2005, trialed in 2006, and has since been commercially deployed in several countries around the globe. The total service area possible for a DVB-H system is limited by the number of transmitter towers that may be economically constructed to enhance the overall coverage area of the service. Non-urban or semi-urban areas have a low population density, which puts a strain on the per capita capital investment that can be made to extend the reach of the service to all parts of the coverage area. In order to extend the reach of the service without additional budgetary strains, a new DVB-SH specification was proposed in 2006, and approved in February 2007 for video broadcast in the S Band using a Geostationary Earth Orbit (GEO) satellite operating in the S band. This specification allows for broadcast of multimedia content directly to a handset using a combination of a satellite and terrestrial wireless link. 1.2 Motivation Several research and development efforts have been launched to study the affects of portable and mobile receivers while receiving and decoding DVB-T signals. A summary of the affects of portability and mobility on signal distortion of DVB-T signals is indicative of the motivation behind the development of a separate standard for handheld devices. Table 1 shows performance classified as “mobile” and “portable”, for long distance/high velocity handheld devices, such as cellular phones, palmtops and PDAs, and short distance/low to medium velocity appliances such as laptops and portable TV sets, respectively. Three different types of potential sources of impairments are outlined in the first column, with specific phenomenon listed in the second column of the table. Possible observed effects and/or specific methods to mitigate its effects are listed in the third and fourth columns. For instance, Mobile receivers suffer from deep fades and flat fading phenomenon under Rayleigh channel conditions; consequently, a large number of (if not all) subcarriers within a single OFDM symbol may be attenuated below the noise floor (i.e. permanently irrecoverable). Likewise, Doppler spread unique to high speed mobile receivers cause
7� HSC
the FFT coefficients to spread across the frequency range (called “spreading”), which contributes to additional noise in the form of Inter Carrier Interference (ICI). Various signal processing methods have been employed throughout the newer DVB-H and DVB-SH systems to mitigate the effects of such impairments.
Mobile Portable
Rayle
igh
Low coherence bandwidth
Channel Equalization (Estimation of channel transfer function)
Deep fades, flat fading
Loss of all or a large number of sub-carriers in an OFDM symbol
Tim
e v
aria
nce
of th
e c
han
ne
l
Low temporal coherence Track variations of the
channel transfer function
People moving in a room can cause picture loss due to deep or flat fading
Low spatial coherence
Accurate placement of receiver needed to avoid losing all or a large number of carriers
Loss of orthogonality due to Doppler spread
FFT leakage causes additional (ICI) noise
Table 1: Performance impairments for mobile and portable devices receiving a DVB-T signal
2.0 SYSTEM ARCHITECTURE The DVB-H system is designed to provide cellular coverage over urban and rural settings using a set of network transmitters which are connected to the back haul networks using high speed data links such as WiMAX or ATM. Being a broadcast network, the data link is unidirectional, from the base transmitter to the hand held device. The coverage area of each transmitter and the overall capacity (in terms of total bandwidth and size of population serviced) is decided by a wide range of factors, including OFDM modulation mode, channel bandwidth and Guard Interval period (cf. 2.1 below). 2.1 DVB-H The transmission data chain of the DVB-H is a very simple unidirectional (“forward only”) link. Data in the form of MPEG2 Transport is encapsulated into IP packets in the network layer. After virtual interleaving and 64-bit Reed-Solomon (RS64) Forward Error Correction (FEC) in the Logical Link Control/Media Access Control (LLC/MAC) layer (cf. 3.1.2 below), the data stream is sliced into Transport Stream (TS) packets and fed to the Physical layer. The physical layer provides a double layer of error protection (RS16 and Convolutional coding) along with randomizing (to maintain bit synchronization) and bit and symbol level frequency domain interleaving (in order to mitigate frequency selective fading). The bit stream is mapped into a set of mutually orthogonal sub carriers within an OFDM symbol. A Guard Interval (GI) separating consecutive OFDM symbols and Cyclic Prefixing (CP) within each GI period provide further protection against multipath and Doppler interference. The OFDM symbols are encapsulated into OFDM Frames (68 OFDM symbols), Super Frames (7 OFDM frames) and Mega Frames to provide synchronization reference points for the receiver. Data from a single transmission channel (“service”) is transmitted at regular intervals in temporal blocks called “Bursts” (sometimes called “Time Slices”). Data bursts from different services are time-multiplexed in a sequential fashion, and transmitted over a specific frequency band. Legacy MPEG2 streams may also be transmitted in a continuous fashion (without time slicing) over the same link, but at a different frequency band. In this way, the same infrastructure may be used to co-locate DVB-H and DVB-T transmissions, thereby saving on costs.
8� HSC
The DVB-H receiver is time synchronized to the specific service the user wishes to avail of at the particular time. In other words, just prior to the time instant when the time slice for the current channel is expected to arrive, the receiver wakes up the demodulator unit and begins receiving the bit stream of the current time slice. Once transmission of the current time slice is completed, the demodulator goes back to “sleep” mode, till the next time slice (for the current service) is expected to arrive. Since time slicing is a means to achieve both better power consumption characteristics as well as maintain optimal bandwidth usage, the time interval between any two time slices of the same service need not be constant. The virtual time interleaving and RS64 coding provides an additional level of FEC coding. The IP packets from the network layer are buffered into a 2 Mbits frame buffer and encapsulated into Multi-Protocol Encapsulation (MPE) datagrams (called MPE “sections”). The combination of error coding and encapsulation is known as MPE-FEC, which is a defining characteristic of DVB-H. FEC remains an optional feature to enhance the Bit Error Rate (BER) of the overall data link. When FEC is applied, the data sections are labeled MPE-FEC sections.
9� HSC
10� HSC
11� HSC
The value of the time lag between any two slices of the same service is denoted as delta-t (t). Each time slice corresponds to the data buffered within one MPE-FEC frame, which is made up of several MPE and/or MPE-FEC sections of the same service. The header of each MPE and/or MPE-FEC section has upto 5 bytes to signal the earliest possible (relative) time when the next burst (for the same service) may start. Such a relative timing system suffers from jitter (known as Delta-t Jitter, Dj). The effects of this form of jitter may be mitigated by decreasing both the absolute value as well as the accuracy of D j. Another parameter which is used to synchronize the “wake up” and “sleep” time of the DVB-H receiver is the Maximum Burst Duration (MBD), which is signaled within each time-sliced elementary stream (ES). Thus, the receiver may not turn on before T1, and shall not stay on longer than T2 = T1 + MBD, where T1
is signaled by the t values of the previous burst, and MBD is signaled for each individual ES in the transmission. Under poor receiver conditions, this information is used to determine the time when a burst is expected to end, and hence, the time when the receiver has to be turned off. Assuming a burst size (Bs) of 2 Mbits (the size of all MPE and/or MPE-FEC section payloads), a burst bandwidth (Bb) of 15 Mbps (the size of all transport packets in the current packets), the maximum duration of the burst (Bd, from the beginning of the first MPE/MPE-FEC section to the end of the last within the current burst) as 140 ms, the constant bandwidth (Cb, without time slicing) as 350 kbps, the synchronization time as 250 ms and the ECM Synchronization Time, St (cf. 9.1, 9.2.3 and 9.2.7 in [8] for details) as 20 ms, the total power saving (%age) for various values of Delta-t jitter (Dj) is plotted in Figure 3 (cf. section 9.2.3 of [8]). It is clearly observed that varying the accuracy of the value of delta-t in order to control jitter does not have a noticeable effect on the gained power saving. Since the main function of a DVB-H system is the broadcast of multimedia (video and/or audio) data in the form of a service, additional data in the form of Electronic Service and Program Guides need to be relayed to every receiver at regular intervals. Such information is relayed in unique MPE encapsulated data sections called Service Information (SI) streams. Likewise, the bouquet of all channels/programs being transmitted, their specific time slice and frequency allocation information, Program ID (PID) values for each video and audio stream within each program and similar information are encapsulated within unique MPE sections called Program Service Information (PSI) sections. Both SI and PSI sections constitute overhead to the data bandwidth in the LLC/MAC layer (cf. Figure 1). 2.2 DVB-SH The DVB-SH standard provides for universal coverage for digital video broadcast by combining the terrestrial coverage, a legacy of the earlier DVB-H standard, with a satellite based coverage network. The two components are called the Satellite Component (SC) and the Complementary Ground Coverage (CGC) component. As the name suggests, the terrestrial network acts as a complementary “gap filler” to the primary signal obtained from the GEO satellite. Content is relayed from the content provider through a Broadcast Distribution Network (BDN), which is typically based on xDSL, WiMAX or any such high capacity back-haul network. The signal may reach the handheld receiver through four distinct paths. An SC signal may be directly broadcast to the mobile hand-held receiver from the satellite, and is widely used in rural settings where signal reach is restricted due to the prohibitive cost of setting up cellular transmission towers to cover the entire service area. A CGC signal may also be broadcast from either Type A, B or C transmitters (cf. Figure 4). Type A transmitters (cf. TR(a) in Figure 4) are “fixed” broadcast infrastructure transmitters, which may be co-located with DVB-H and/or other cellular transmitters, and local content insertion is possible in such a scenario. This type of arrangement is suitable for urban scenarios where satellite coverage may be unsuitable due to an extremely high noise floor, overloading of capacity or “urban canyon” effect.
12� HSC
Type B transmitters (cf. TR(b) in Figure 4) act as passive relay devices, or perform frequency conversion before relaying the DVB-SH signal to the receiver. These types of transmitters act as small range gap fillers to enhance the reach of the satellite signal in indoor environments. Type C transmitters (cf. TR(c) in Figure 4) are “mobile” broadcast infrastructure transmitters, creating a “mobile complementary infrastructure”. 2.3 OFDM in satellite links 2.3.1 Line of Sight (LOS) In the bands at which most DVB-H and DVB-SH systems are expected to operate (> 2 GHz), the signal propagation properties of Radio Frequency (RF) electromagnetic waves preclude the use of Line-of-Sight (LOS) applications such as a direct satellite link. 2.3.2 Amplifier non-linearities and Output Back-Off (OBO) In the extremely unlikely situation when all the sub-carriers (each with a peak power of 1W) within a single OFDM symbol are transmitted with maximum amplitude simultaneously, the maximum value of power of the transmitted symbol is achieved. For a 2k mode symbol, this “peak power” level is approximately 33 dB above its average value – in other words, the Peak to Average Power Ratio (PAPR) of the transmitted signal is about 33 dB. A more probable situation occurs when the transmitted signal achieves a RMS value of PAPR, which in this case is approximately half the “maximum” possible PAPR, or 16 dB [3]. Most satellite system designs employ highly non-linear High Power Amplifiers (HPA) in the front end of the receivers. The large amplitude variation introduces an unacceptably high amount of intermodulation levels due to non-linear amplification, which may be manifested as an amplifier Output Back-Off (OBO) of about 3 dB. Modulation schemes such as QPSK, 16QAM and 64QAM are generally robust enough to operate under very high spectral re-growth levels. However, too much of distortion in the carrier envelope could destroy the high spectral efficiency of the OFDM symbol. A typical biased Class A power amplifier with output power levels of 7 to 10 dB can maintain the intermodulation levels at an acceptable level of approximately -40 dBc [3]. Linear distortion introduced in the downlink path of a satellite signal may be attributed to the Input Multiplexer (IMUX) and Output Multiplexer (OMUX) units of the satellite transponder, as well as the multipath effects while transmitting to several repeaters at ground level. 2.3.3 Suitable remedies possible It is generally assumed that the uplink from the terrestrial feeder link and the satellite is ideal. Both linear and non-linear distortion within the satellite repeater data chain and the downlink from the satellite transponder to the handheld device may be mitigated by using appropriate pre-distortion techniques ([12] and [13]) and application of appropriate channel coding, mostly in the form of turbo codes. The pre-distortion may be in the form of a passive Look-Up-Table (LUT) acting on individual samples to compensate for the constellation warping introduced by the phase distortion introduced by the HPA characteristic, as well to reduce the variance of the clusters of constellation points. As documented above, COFDM transmission is generally considered unsuitable, if not downright impractical for transmission on long haul satellite links. Hence, the satellite broadcast link is available in both COFDM as well as classical Time Division Multiplex (TDM) mode. This gives rise to two modes of operation for the DVB-SH system:
13� HSC
1. DVB-SH-A: This mode employs COFDM over both the primary satellite and secondary terrestrial
“gap filler” network. 2. DVB-SH-B: This mode employs COFDM in the terrestrial link, and reverts to classical TDM in the
satellite link. Issues related to channel impairments that are unique to satellite links may be addressed by introducing flexibility in the nature of interleaving in the physical and LLC/MAC layer. Based on this, DVB-SH receivers are classified as:
● Class 1 Receivers: The DVB-SH Mega Frame is in the order of one DVB-H time slice, and provides “short” physical layer protection. This is complemented with a longer LLC/MAC layer interleaver.
● Class 2 Receivers: The DVB-SH Mega Frame supports long physical layer protection (in the order of several DVB-H bursts or time slices). The quality of protection in the service layer may be tuned to the quality of service expected.
Class 1 receivers are able to cope with rather short-term fading and blockage, such as being obstructed by a tall building, overpass or bridge, trees and other permanent obstacles which may be encountered by the receiver during motion. These obstacles may be experienced for not more than a few milliseconds to about 1 second (depending on the relative velocity of the receiver terminal). Additional fading mitigation is provided for in the LLC/MAC layer. DVB-SH-B specifies an additional Class 2 receiver, which provides long term fading mitigation of upto 10 seconds. It is upto the service and network operators to allocate the protection between the different layers, depending on the QoS expected, service categories and commercialized class of receivers in the market. 2.4 SFN networks vs. MFN networks A unique feature of the DVB-H (as well as DVB-SH) system is the capability to synchronize the entire terrestrial broadcast network (CGC in case of DVB-SH) to one single carrier frequency. Such networks are called Single Frequency Networks (SFN). Traditional broadcast networks operate on the principle of dividing the entire coverage area into small cells, each serviced by a separate broadcast antenna and transmitter system. Each cell operates on a separate frequency band, thereby preventing inter-cell interference. Frequency reuse is possible by arranging cells using the same frequency band as far away from each other, without abutting each other. However, this poses soft and hard hand-off issues, time slice synchronization problems, as well as exorbitant capital investment in multiple spectrum bands. In such Multi Frequency Networks (MFN), the receiver has to continuously tune to all possible frequency bands in which the current service provider is providing broadcast services, and lock on to the cell and frequency band with the strongest signal strength. Moreover, the receiver has to operate on all possible bands during the “off period” between time slices, in case the current program switches to another carrier frequency during hand-off between cells. This nullifies any advantage of power efficiency that may have been accrued due to the unique time slicing nature of transmission in DVB-H. Single Frequency Networks on the other hand allow all transmitters of a service provider covering a given service area to operate at a single carrier frequency. Every receiver within the service area will need to be tuned to a single fixed band, irrespective of the specific transmitter it might be receiving its primary signal from. The receiver will now not only receive a primary signal (either LoS or with minimal multipath) from one transmitter, but also receive delayed (LoS and/or multipath delayed) versions of the same signal from
14� HSC
all other receivers in the network. Ofcourse, the farther the transmitter is from the receiver, more shall be the attenuation and delay in the reception of the “secondary” versions of the signal. SFN networks operate on three main principles: “Each transmitter in the SFN network shall radiate”
● at the same frequency
● at the same time
● the same data bits Such a system requires sophisticated synchronization methodologies which synchronize the carrier center frequency and the COFDM sub-carrier frequency of all transmitters in the network at any given point of time. Time synchronization of all the transmitters is of utmost importance, since all the (delayed) versions of the signal must reach the receiver within a time window equivalent to the total time interval (useful sampling time + GI) between consecutive OFDM symbols. This helps in confining the maximum LoS and/or multipath version of the signal to less than the GI period to which the system specification has been set. A Primary Distribution Network (PDN) carries the bouquet of all program data to the multiplexer, from whence it is radiated from the transmitter (cf. Fig. 5). A Network Transit Delay (NTD) and Network Padding Delay (NPD) need to be accounted for while the multiplex data is being processed within the PDN. To compensate for this delay, a “negative” Network Compensation Delay (NCD) needs to be introduced in the PDN. Along the same lines, each transmitter introduces a Tx Processing Delay (TPD) and a Tx Offset Delay (TOD), both of which are compensated using a Tx Compensation Delay (TCD). Both the compensation delays should be in synchronized in the PDN as well as every transmitted in the system such that the same data is transmitted from each network antenna at exactly the same time. The time synchronization to compensate for PDN and Tx delays is obtained using a unique time stamp (called Synchronization Time Stamp, or STS) generated by a set of GPS units, which derive their synchronization parameters from a common platform (US GPS system, for example). 2.5 Synchronization of SC and CGC signals in a SFN network SFN network architecture poses additional challenges with regards to the synchronization of the received waveforms at the mobile receiver. A single copy of the same signal has to be received from one or several of the terrestrial components of the DVB-SH system “at the same time”, and depending on the resultant (constructive or destructive) interference, would result in boosting or attenuation of the recived signal. The challenges faced during synchronization depend on the type, range, power level and relative velocity of the complementary ground based transmitter. Synchronization mechanisms differ only slightly between DVB-SH-A and DVB-SH-B modes. Consider the example of a Mobile Satellite System (MSS) operating in a 15 MHz bandwidth in the lower S band. These could be further subdivided into three 5 MHz bands to be allocated to 3 different spot beams 2.6 Synchronization in DVB-SH-A In the DVB-SH-A system, each transmitter in an SFN system is tuned to exactly the same 5 MHz band as the satellite transponder, and transmits exactly the same bit stream synchronously to the mobile receiver in the coverage area. The synchronization is achieved using a SHIP packet, which allows the “slave” CGC transmitters to transmit exactly the same bits as the SC transmitter. Since the relative delay between individual CGC transmitters and the receiver is in direct relation to the distance from the receiver, the value of the propagation delay of each antenna sub-system within the SFN shall be different.
15� HSC
The broadcast feed network (cf. Fig. 4) introduces the appropriate compensation delay values through the individual SHIP packets. 2.7 Synchronization in DVB-SH-B In the DVB-SH-B system, the band allocated for the TDM based SC is allocated a different 5 MHz band than the CGC. The receiver can make use of the same digital symbols received after demodulating the OFDM as well as the TDM waveforms, and hence the same synchronization technique used for DVB-SH-A is applicable. However, because of the relatively higher SNR of terrestrial transmissions, local content insertion is possible within the same waveform in the case of DVB-SH-B CGC. Synchronization issues are especially complex when data from TR(a) (fixed infrastructure) and TR(c) (mobile infrastructure) class transmitters need to be sync-ed with the SC signal. In the case of the TR(b) (personal gap fillers) class transmitters, the size of the cell served by the personal gap filler is small enough that the delay due to (optional) frequency conversion, power boosting and retransmission may be ignored. 2.8 TR(a) and TR(c) class transmitters Fixed and mobile infrastructure CGC transmitters source their signal directly from the Content Provider (CP), usually through high speed IP based back haul systems such as ATM or WIMAX. Similarly, the DVB-SH feed network receives its copy of the signal directly from the CP. The SC component of the system has to be transmitted via uplink and downlink to the mobile receiver. This entails an additional delay of 2*(2.6788925/30) = 0.245 seconds per symbol in comparison to the direct transmission from the CGC transmitters. Assuming a median distance of about 4 km from a transmitter to the mobile receiver (cf. section 6.1.2.6 for a discussion on maximum detection range), the
delay introduced by the CGC signal (which is in the range of approx. 13 s), may be safely ignored. Thus, in order to ensure proper synchronization between the SC and CGC signals, an additional delay of (approx.) 0.25 seconds must be introduced at the SFN synchronization system. This calculation ignores the minute delay introduced by the terrestrial transmission, but includes delays caused by internal signal processing within the transmitter (power boosting, frequency conversion, error correction routines, and occasionally, LCI). These delays are manifest in the form of NCD and TCD at the backhaul and transmitter antenna front ends, in order to ensure proper synchronization. 2.9 TR(b) class transmitters Personal gap fillers are typically used to enhance the power level of the SC, and retransmitted at the same or different frequency over short distances in indoor environments. LCI is not possible in this case, and hence the additional delay due to local content insertion may be ignored. The uplink and downlink transit times for both the direct SC and the TR(b) CGC signals are exactly the same (ignoring the minute differences in slant path lengths to the gap filler and mobile receiver antennas). In addition to this, the processing time and transit time from the gap filler antenna to the receiver antenna is typically in the order of several milliseconds, and hence may be ignored in comparison to the 0.245 seconds transit time for the SC. In the case of TR(b) transmitters, NCD may be assumed to be zero (since there is no backhaul network), and hence the entire temporal compensation is performed by the TCD in the Tx antenna front end.
16� HSC
17� HSC
18� HSC
3.0 ARCHITECTURAL CONSTITUENTS 3.1 Constituents of the DVB-H system The DVB-H and DVB-SH systems are a superset of (and completely backward compatible with) the legacy DVB-T system. Both the DVB-H and the DVB-SH standards specify the PHY (baseband DSP) and LLC/MAC layer operations of the protocol stack. Unlike the terrestrial broadcast specification, which relied heavily on MPEG-2 Transport Stream (MPEG2 TS; labeled as “TS” in the figure 1) to carry encoded multimedia content, the newer systems incorporate an IP based transport scheme into its stack. This is helpful in a seamless integration of the PHY and MAC layers onto a regular IP based protocol stack in most internet based applications. Much of the innovation to provide “mobile” video has been specified in the LLC/MAC layers of the specification. A few minor additions and modifications to the PHY layer provide added flexibility to the coverage and frequency planning process. 3.1.1 DVB-H and DVB-SH embedded within DVB-T Fig. 1 shows the protocol stack of the DVB-H and DVB-SH systems as a modified addendum to the original DVB-T scheme. Specific features in the DVB-H MAC/LLC specification include MPE-FEC error correction system and the Time Slicing scheme to conserve battery life. The specific improvements in the physical layer include Transmission Parameter Signaling (TPS) signaling modifications and an additional 4K FFT mode, amongst others. In the case of DVB-SH, TPS bit fields have been generally left unchanged for purposes of backward compatibility. The OSI specification layers covered in figure 1 alongside include the PHY, LLC/MAC, IP, Service and Application layers, from bottom up. For e.g., the figure shows the DVB-SH (denoted by its legacy nomenclature – DVB-SSP, or DVB – Satellite Service to Portable devices) and DVB-H systems to be peer specifications in the LLC/MAC layer. Both accept IP data from the network layer and encapsulate them into standardized MPE sections. The physical layer of the DVB-H system is adapted from the legacy DVB-T with some minor additions and modifications. On the other hand, the PHY layer specification of the DVB-SH system is a combination of modules from the legacy DVB-S and DVB-T systems. This is especially noteworthy considering that the DVB-SH system operates in 2 modes (cf. 2.3.3 above for details). 3.1.2 The LLC/MAC layer The DVB-T specification provides for a multicast unidirectional path available to all receivers which are tuned to the proper carrier frequency; hence, concurrent access to the medium is a non-issue. However, IP based content needs to be modified to be carried over a common “container” specification which is backward compatible with DVB-T. The handheld specification achieves this by employing a Multi-Protocol Encapsulation (MPE) mechanism to map IP data onto MPEG2-TS packets. An additional level of error protection is provided using the Reed Solomon (255, 191, 64) Forward Error Correcting (FEC) coding scheme (succinctly called RS64). The MPE module comprises an Application Data Table (ADT) and Reed-Solomon Data Table (RSDT). Together they simulate a “virtual interleaving” process which ensures that a noise burst does not affect a continuous sequence of bytes in the data stream. The IP packets are streamed into the ADT column wise, and padded with zeros if some space in the 191 columns of the ADT remains unfilled. Each row of the ADT is then RS64 coded to generate a 64 byte parity code in the corresponding row of the RSDT. Each row of the RSDT is 64 bytes wide. The number of rows in the ADT/RSDT may be variably fixed at 256, 512, 768 or 1024.
19� HSC
The IP packets and the 64 byte RS codewords are then encapsulated into data blocks or datagrams called Multi-Protocol Encapsulation (MPE) and MPE-FEC Sections, respectively. Further protection in the form of MPE and MPE-FEC headers and an optional CRC tail is provided. The MPE and MPE-FEC sections are then fed to the Transport Stream (TS) slicer module to generate legacy TS packets, thereby making the data stream syntactically compatible with DVB-T. The handheld specification employs a special “time-sliced” process of Over-the-Air (OTA) transmission; data within one service is episodically transmitted within individual time slices. In this fashion, multiple services (combinations of different programs, or same video content formatted for HDTV and SDTV transmissions, for example) may be multiplexed into the same transmission system. A user may tune into a single channel by locking onto the transmission within one time slice. Once it has been locked onto a single channel, the receiver is synchronized to “wake up” at precisely regular intervals of time, just prior to each impending time slice and receive the bit stream, and then go back to “sleep mode”. The back end systems (DSP, network, session and application layer applications) shall then process the data received within one single time slice (and possibly display the received video data) before it is time to begin receiving the next time slice of the current service. The MPE-FEC specified in [8] is suitable for short duration fading (of upto the duration of a single data burst). However, a typical satellite downlink suffers from much longer fading due to deep and long shadowing, especially in urban canyon environments and very heavy precipitation. To mitigate the effects of such “long duration” fades, it has been recommended that the outer FEC (a Reed-Solomon 64 code with virtual time interleaving similar to the one used in DVB-SH) be extended to data across longer time durations, namely, multiple time slices. The “Extended” MPE-FEC protection provides for additional time diversity, thereby helping in combating deep and long shadowing fading effects. An additional innovation introduced in the MAC layer of the DVB-SH system is the SHIP (cf. Section 2.5) packet. This is a ISO/IEC 13818 (MPEG2) compliant TS packet which is 188 bytes long, and has some pre-defined header and payload fields. Each SH frame shall contain exactly one SHIP packet. The primary purpose of the SHIP packet is to synchronize the data sequence being transmitted from all the CGC and SC transmitters within the same SFN. Certain optional fields in the SHIP packet which are used to achieve this include:
synchronization_time_stamp
Transmitter Time Offset Function
Service Synchronization Function Besides synchronization functions, the SHIP packet also transmits information about the bandwidth allocated for the particular transmission, BCH code for error control of the TPS signaling stream, cell identifier value and DVB-SH super frame number.
20� HSC
3.1.3 Encapsulation and packetization in the DLC layer The basic unit of data which is fed as the input to the LLC/MAC layer of the transmitter is the IP datagram. Each datagram is of a fixed size (1024 bytes). After generation of the (optional) RS64 codewords (cf. 3.1.2 above), the IP packets are encapsulated into individual container objects known as MPE sections, with one MPE section used to contain only one IP packet. A 12 byte header and a 4 byte Cyclic Redundancy Check (CRC) code tail are also appended. Similarly, the RS64 FEC codewords are also encapsulated into individual sections (called MPE-FEC sections). The Program Service Information (PSI) and Service Information (SI) data are also likewise encapsulated into sections. Unlike regular data payloads, however, these sections are not transmitted within time slice boundaries. Each section containing PSI or SI data is made up of a “table” of the respective type, with individual control fields (called “descriptors”) duly filled in (cf. [6] and [8] for details on the construction of PSI and SI data). MPE sections payloads are then “sliced” into packets of size 184 bytes and pre-appended with a 4 byte header, together known as “TS Packet”. The header of the TS packet is based on the MPEG2 Part 1 (Systems) specification (cf. [11] for details). Each header contains a Program ID (PID) field, which describes the type of data (SI/PSI or payload) carried by the respective TS packet (cf. [6] for details on all header and descriptor fields). The 188 byte TS packet acts as the input to the PHY layer of the DVB-H/DVB-SH system. 3.1.4 The PHY layer The DVB-T PHY layer provides for an Orthogonal Frequency Division Multiplexing (OFDM) scheme, wherein the entire bandwidth of the channel is subdivided into a fixed number of sub-carriers. Each sub-band carries a single subcarrier signal modulated by the digital bit stream using QPSK, 16 QAM or 64 QAM digital modulation. A single OFDM symbol is represented by a series of slots along the frequency axis (cf. Fig 7). The individual frequencies of the sub-carriers and the spacing between them are chosen in such a way that they are mutually orthogonal to one another; In other words, the sidelobes of each subcarrier do not interfere with the sidelobes of any other subcarrier within the same OFDM symbol. Due to this mutual orthogonality, the coherent demodulation and detection of all subcarriers (and the modulating bits being carried by it) may be performed using a regular N-point Fast Fourier Transform (FFT), without the need for expensive band pass filters to isolate each individual subcarrier. Along the time axis, each OFDM symbol is separated from the next symbol by a Guard Interval (GI). The receiver may receive a symbol either transmitted directly from the transmitter antenna along the Line of Sight (LoS), or multiple copies of the symbol after being reflected by obstructions along the signal path.
The GI is set at such a value so as to account for the maximum propagation delay possible due to multipath in the signal environment under which the system is being modeled. No processing takes place during the GI. All detection and demodulation processes (e.g. FFT) have to be timed so as to occur only during the specific time window when the OFDM symbol is actually being “detected” by the receiver. Variable channel conditions due to changing weather and atmospheric conditions in the signal path lead to
21� HSC
channel impairments which may be modeled as a time varying FIR filter. To mitigate the effects of such a distortion, a Decision Feedback Equalizer (DFE) needs to be inserted in the front end of the demodulator in the receiver unit. Equalization is typically carried out by modeling the instantaneous channel characteristic as an FIR filter, and convolving the instantaneous input to the receiver with the inverse of the DFE FIR impulse response. The DFE is actually executed in the frequency domain after the FFT step at the receiver, thereby allowing the DFE to be implemented using a set of simple multiplicative steps instead of the more complex “multiply-add-shift” step in convolution. Since the signal is received and processed in the discrete time domain, the “linear” convolution used to model the DFE is converted into a “circular” one. This is achieved by duplicating the tail end of each OFDM symbol in the GI preceding the current symbol. This form of pseudo-symmetry is called “Cyclic Prefixing” (CP). The time varying characteristics of the equalizer is aided by the availability of Channel State Information (CSI), which is gleaned at regular intervals during the transmission from Pilot Carriers interspersed at regular intervals in the channel. The pilot carriers do not carry any useful payload data. In the figure above, pilot carriers are shown using dark shaded slots. The number of pilot carriers in a single OFDM symbol is a variable design factor to be accounted for in the system design. More number of pilot carriers provides for robust detection of CSI, but reduces the overall useful bandwidth. Fig. 1 above shows the typical data chain of the PHY layer of the DVB-H system. Two levels of error protection are provided by the outer (Reed Solomon (204, 188, 16)) and inner (Connvolutional – Viterbi) coder. The overall energy of the signal is spread along the time and frequency axis using a scrambler, which simulates a GF(2) scrambling algorithm. The effect of using the scrambler is to prevent the occurrence of long sequences of zeros (binary low symbols) or ones (binary high symbols) which might cause the system to lose bit synchronization.
Two levels of native interleaving at the symbol and bit level are provided for in the DVB-H and DVB-SH systems. These blocks help in spreading individual bits and symbols (sequences of bits) within the same OFDM symbol, thereby mitigating the effect of frequency selective fading within the same OFDM symbol. These interleavers work on a fixed number of bytes of data stored in a contiguous block in memory; the size of the memory block depends on the FFT mode in use, with
the 8K mode employing the single largest block of memory. The smaller 2K and 4K modes employ memory blocks which are a relative fraction of the memory size. During the interleaving process, bits and bytes carried within sub-carriers of the same OFDM symbol are rearranged within the same OFDM symbol (along the frequency axis) according to a fixed pattern. Such interleaving is called Frequency Domain Interleaving (FDI). A special characteristic of mobile networks covering large networks is a form of “long term” fading known as Rayleigh Channel Fading. Rayleigh channels suffer from frequency selective nulls and attenuation which propagate over long periods of time (multiple OFDM symbols). The distortion effects of this phenomenon may be mitigated by implementing an interleaver along the time axis. This is achieved by a Time Domain Interleaver (TDI) called the In-Depth Interleaver (IDI). This is an optional feature specific to
22� HSC
the 2K and 4K mode. Four 2K OFDM symbols or two 4K OFDM symbols (within the same time slice) are arranged in the contiguous memory block described earlier (equivalent to one 8K OFDM symbol), and the data is interleaved along the time axis. This arrangement precludes the use of extra memory while still allowing the transmission of data in time slices. 3.1.5 Issues specific to DVB-SH The physical layer of the DVB-SH system has been derived from the older DVB-H system. As discussed earlier, the standard specifies two modes of operation. In the case of DVB-SH-A, both the SC and CGC employ an OFDM modulation scheme using an additional 1K mode in addition to the legacy 2K and 8K (DVB-T) and 4K (DVB-T) modes. The standard specifies a 3GPP2 Turbo coding error protection module (with puncturing) instead of the RS-16 and Rate Compatible Punctured Convolutional Code (RCPC)-Viterbi decoder combination. The physical layer also employs an additional 1.7 MHz channel specification apart from the original 6 MHz, 7 MHz and 8 MHz channel specification (for DVB-T and DVB-H) and 5 MHz (for DVB-H only). The SC and CGC of DVB-SH-A employ either a QPSK or a 16QAM constellation scheme, thereby limiting the number of bits per sub-carrier in an OFDM symbol to upto 4 (against a maximum of 6 bits/sub-carrier for DVB-H). The CGC mode of DVB-SH-B resembles the CGC of DVB-SH-A described above. However, the SC of DVB-SH-B employs a Single Carrier TDM unit (instead of OFDM) with QPSK, 8PSK or 16APSK modulation (cf. sections 2.3.2 and 6 respectively on why OFDM and QAM are unsuitable for satellite bound payloads), along with a randomizer (scrambler) for energy dispersal. The signals are transmitted in pre-defined Physical Layer (PL) slots. These system components are identical to the one specified in the DVB-S2 system specification. 3.1.6 Interleaving Two different types of interleaving are performed in the PHY layer of the DVB-H and DVB-SH in order to enhance the resistance of the transmitted waveform against short term fading and medium term shadowing effects in both the SC and CGC. A bit wise interleaver is used for both OFDM (SC or CGC) as well as TDM (SC only) waveforms in order to provide the bulk of the channel diversity required to mitigate the effects of fading and shadowing. OFDM symbols in the SC and the CGC also employ a symbol level interleaver. The bit interleaver is operated on output blocks of the Turbo encoder of size 1146 bits (for signaling field) or 12282 bits (for payload). The bits are shuffled within the boundary of the input block using a standardized modulo n operation. For reverse compatibility with DVB-T frames, the payload bits are punctured after bit interleaving. Every 128 bits block is modified by transmitting the first 126 bits, thereby ignoring the last 2 bits in the block. This form of Rate Adaptation ensures that the frame structures for DVB-SH and DVB-T are mapped. Each Interleaving Unit (IU) of size 126 bits is used as a cell during the ensuing Time Interleaving. This interleaver is identical to the one specified in the legacy DVB-T specification. The depth of each of the 48 cells (L(0) to L(47)) is specified by the TPS stream (for OFDM carriers) or Signaling Field (SF) (for TDM carriers). Symbol interleaving is performed only in OFDM transmission (in both the SC and CGC). It involves the sparse mapping of a single bit word of length „v‟ onto the active carriers of an OFDM symbol (756 in 1K mode, 1512 in 2K mode, 3024 in 4K mode or 6048 in 8K mode). A complex permutation function, H(w) ensures that each „v‟ bit word is spread evenly across several sub carriers within a single OFDM symbol.
23� HSC
Symbol interleaving provides a suitable antidote to frequency selective fading within a single OFDM symbol. This form of fading affects only selective sub-carriers (frequency components) in an OFDM symbol, and this form of interleaving ensures that no two consecutive payload bits are distorted or completely lost when neighboring OFDM sub-carriers are mutilated due to fading. Irrespective of the actual duration of the fading phenomenon, symbol interleaving ensures that no two consecutive payload bits are transmitted within the same or even neighboring sub-carriers. 3.1.7 The Holy Grail of mobile reception The situation is further complicated during mobile reception, since the receiver now not only has to contend with a multiplicity of received echoes with variable time delays (due to multi-path), but also include frequency shift introduced due to the Doppler Effect into the calculations. Echoes affected by Doppler frequency shifts are perceived as noise contributing to the Inter Channel Interference (ICI) component of the channel characteristic. Signal processing algorithms designed to mitigate the effects of Doppler shift make use of channel estimation techniques (gleaned from the periodic pilot carriers) to reduce the ICI noise below the level at which the orthogonality of the subcarriers is lost and demodulation is made impossible. 3.1.8 Modulation of the baseband signal Digital modulation is the technique by which digital information is piggybacked onto an analog carrier wave signal such as an RF signal. Both DVB-H and DVB-SH systems employ multiple modes of digital modulation which provide varying levels of robustness and channel capacity. Currently, the standards specify 4QAM (also known as 4PSK or QPSK), 16QAM and 64QAM as the three possible modes available. In fact, these modes are also available in the legacy DVB-T system. As is evident from Fig. 1, the channel capacity of a particular modulation scheme is measured by the number of points in a constellation (in a binary system, a 16QAM modulation scheme would allow 4 (2
4 =
16) bits per sub-carrier, for example); more the number of points, the higher the channel capacity and the bit rate. However, by packing more points within the same signal space, the constellation points are brought closer to one another, thereby allowing even minor noise impulses to allow erroneous detection of the received symbol as a different point in the constellation. This has the effect of reducing the measure of robustness of the channel (64QAM tries to fit 64 points within the same signal space where QPSK would fit in only 4 points, and hence is 16 times less robust). The robustness of the system may be increased by moving the group of points in each quadrant further away from the cardinal axes; this has the effect of increasing the distance between points in two different quadrants, and thereby reduces the probability of a received symbol being erroneously detected in an altogether another quadrant. The measure of separation between quadrants is
measured by a quantity labeled . The innovation introduced into the handheld specification is called Hierarchical Modulation. This allows the embedding of 2 lower rate (but comparatively more robust) modulation schemes into a single higher rate (but comparatively less robust) scheme. For e.g., a 64QAM modulation scheme may embed 6 bits onto each sub-carrier of an OFDM symbol (the figure above shows one such symbol representing the bit sequence “110100”). However, by segregating these 6 bits into sets of 2 (“11”) and 4 (“0100”) bits, we are able to provide 2 separate data
24� HSC
D = V * ( rf / C ) * cos ( ) Where,
D : Frequency shift due to Doppler effect V : Relative velocity of receiver
rf : Center of carrier frequency C : Speed of light in vacuum (3 x 10
8 m/s)
Angle of approach
Johann Andreas Christian DOPPLER
streams modulated using QPSK and 16 QAM, respectively. The white dots represent one of four symbols of the QPSK scheme – it carries 2 bits per symbol (a comparatively low bit rate, but the larger separation between each of the white dots denotes a better robustness). The cyan dots in each quadrant represent a 16QAM scheme – this carries four times more data, but because of a closer spacing, is more susceptible to noise, and hence is less robust. Together, the 4 sets of 16QAM symbols constitute a 64 point 64QAM constellation. The former bit stream is labeled “High Priority” (HP) – “11” - and has a lower bit rate but is more robust (QPSK is the most robust modulation scheme in use in DVB). The latter stream is labeled “Low Priority” (LP) – “0100” - and has a higher bit rate with less amount of robustness (in comparison with QPSK). However, the combined bit rate of the QPSK-16QAM combination is exactly the same as the (least robust) 64QAM scheme. 3.1.9 Advantages of Hierarchical Modulation – “Simulcast” By using a hierarchical modulation scheme, we are able to achieve the highest possible bit rate (of 64QAM), while still maintaining robustness within acceptable levels (of up to 16QAM). Since two separate and independent data streams (labeled “HP” and “LP”) are being transmitted, the additional header/trailer bytes of the second stream may act as overhead to slightly reduce the overall bit rate. Using this scheme of modulation, two independent programs may be “simulcast” to two sets of consumers (with differing coverage plans) within the same service area. In another scenario, the same video program may be broadcast with varying bit rates and levels of robustness in order to provide coverage to abutting rural/low density as well as urban/heavily populated service areas. Similarly, it may be possible to multiplex a full DVB-H service over the HP stream along with a legacy DTV service over the LP stream for static users who do not need a high level of robustness. In a more innovative and sophisticated scenario, both HDTV (LP; high bit rate and lower robustness; larger coverage area but at a premium price markup) and SDTV (HP; lower bit rate and higher robustness; smaller coverage area but at the normal price range) versions of the same broadcast content could be broadcast to consumers (with different coverage plans) within the same service area. LP streams may compensate for the inherent lack of robustness by employing greater levels of redundancy in the convolutional and RS coding in the PHY layer, in comparison to the HP streams.
4.0 SPECTRUM ALLOCATION AND RELATED BANDWIDTH ISSUES DVB-H and DVB-SH is designed to work on the so-called “traditional” broadcast frequencies. These include VHF Band III (174 MHz to 230 MHz), UHF Band IV (470 to 598 MHz), UHF Band V (598 MHz to 862 MHz), L Band (1 to 2 GHz) and S Band (2 to 4 GHz). However, certain constraints come in the way of full deployment of systems in these bands, thereby compelling the search for alternative (and higher) transmission bands for both DVB-H and DVB-SH.
VHF Band IV: This band has exceptionally good propagation, attenuation (within structures) and Doppler characteristics. However, the typically long wavelength (> 1 m) of these waves forces the receiver antennae to be large in dimensions, making integration onto a mobile device
25� HSC
extremely difficult except in the case of receivers built on vehicles such as cars and trains. Systems for hand-held devices such as cellular phones and PDAs are impossible to build in the band.
UHF Bands IV/V: These bands (with the exception of the upper reaches of UHF Band V, which is already allocated to GSM850/900) have acceptable levels of propagation and attenuation, offer fairly large coverage and allow speeds of upto 250 – 500 kmph on account of Doppler shift. However, these two bands are heavily congested with traditional analog broadcast TV and DTTV transmissions.
Such constraints could compel broadcast to be moved to more non-traditional bands such as the L and S bands. However, according to the Doppler frequency shift formula, the deviation from the center (carrier)
frequency due to relative motion is in direct proportion to the RF center frequency, rf of the band and the
relative velocity, V. Hence to maintain a constant (maximal) frequency deviation, D, the permissible RF transmission frequency may need to be reduced (to VHF bands) to allow for higher speeds (upto Mach 1.0). A simplified mobile performance chart measures the maximum velocity allowed under operation in different bands of the radio spectrum. The data has been accumulated assuming the TU6 model with a specific set of parameters (channel bandwidth: 5 MHz, 2K subcarriers per OFDM symbol (mode), GI: 1/8, 16 QAM constellation, 2/3 inner code rate). The DVB-T bit rate is pegged at 9.2 Mbps, while the corresponding bit rate for DVB-H is 6.9 Mbps (assuming a MPE-FEC code rate of ¾). Fig. 9 shows that in order to maintain a minimum bit rate of 7 Mbps for mobile applications, transmitter coverage area and mobile speed decreases even as the channel carrier frequency increases. For e.g., portable/mobile receivers for pedestrians and buses can work at S band (2.4 to 3.9 MHz), or in cars on highways covered with an L band (1.452 to 1.492 MHz) or in a train covered with a UHF-V signal (470 to 862 MHz) (although mobile receivers in the Shinkansen, TGV and Shanghai transrapid systems may have to tune down to the UHF-IV band) or even to a VHF signal in a Mach 1.0 aircraft.
26� HSC
Fig. 9 4.1 Frequency band selection for the DVB-SH system In the current system setup, it is assumed that the earth station is at a fixed point on the surface of the earth. If the GEO satellite traverses a perfectly circular orbit along the equatorial plane, the sub-satellite
point remains fixed at the same longitude and latitude (0) throughout the journey of the satellite around
earth. In such a case, the sub-satellite point may be viewed to be at a state of perpetual rest with respect to the earth station. Since the sub-satellite point is always directly below the satellite (nadir), it is always at rest with respect to the satellite. Thus, the earth station is in a state of rest “relatively speaking” with respect to the satellite, and hence the effects of Doppler may be safely ignored. However, in most GEO systems deployed for broadcast purposes, the inclination of the satellite orbit is not always maintained at zero, and the shape of the orbit is elliptical rather than perfectly circular. At the apogee, the satellite‟s instantaneous velocity and distance from the nadir may be slightly greater than their typical average values, thereby leading to two effects:
The relative linear velocity between the satellite and the earth station leads to a slight Doppler shift in the carrier frequency (cf. 4.1.1 for details).
The deviation of the total path length from the average value leads to a change in the total path
delay (cf. 7.2 for details).
27� HSC
4.2 Doppler considerations
During the first five years of the satellite‟s life, the orbit is maintained at the equatorial plane +/- 0.05.
After this period, latitude correction ceases, and the inclination builds at an average rate of 0.8 per year.
The relative velocity of the satellite w.r.t the earth station at the apogee (as well as perigee) is given by the empirical formula [16]:
where,
D = Apogee distance of the satellite from the earth station; this distance is computed from the
geometry of the orbital plane
H = Zenith height of the satellite (I have assumed a zenith height of 36786 km)
In the above formulation, D may be viewed as the slant distance from the earth station and the satellite,
and is a function of the instantaneous angle of inclination, (cf. 6.1.1.2.1 for details). Assuming the
earth‟s radius, RE to be 6378.1 km, and a maximum inclination, of 5, the mean relative velocity, V is
4.76 km. This translates to a Doppler shift of (approx.) 10 Hz. This shift increases non-linearly as the
inclination of the orbit increases, and may reach a value of 75 Hz at = 15. In the CGC component of the DVB-SH system, the effects of Doppler are seen to be similar to the classical DVB-H case. Hence, selection of the frequency band for optimal operation would be based on criteria similar to those discussed in 4.0.
5.0 MOBILE MULTICAST/BROADCAST SERVICE (MBMS) 5.1 Multicast IP Service A multicast IP service is a unidirectional Point-to-MultiPoint (P2MP) IP DataCasting (IPDC) service in which data is transmitted from a single source to a multicast group in the associated multicast service area. Users who wish to avail of this service need to have subscribed to the relevant service as well as have joined the multicast group associated with the relevant service on a session basis. The MBMS architecture, as standardized by the 3
rd Generation Partnership Project (3GPP), has been
designed to be an extension of the existing GSM/GPRS/EDGE/UMTS (Rel. 6) network design. More explicitly, the radio resource is shared “in an efficient manner” between traditional GSM/UMTS voice and data services on one hand, and MBMS broadcast/multicast services, on the other. A back-end Broadcast/Multicast Service Center (BM-SC) to provide all relevant end-to-end IP services is an additional component of the traditional GSM/UMTS network architecture. MBMS has been designed for light video and audio broadcast/multicast, although it may be easily scaled up for video streaming applications. Data rates supported by the MBMS architecture depend on the actual radio access network technology (UMTS Terrestrial Radio Access Network (UTRAN) employing Wideband CDMA (WCDMA) or GSM EDGE Radio Access Network (GERAN) employing Enhanced Data rates for GSM Evolution (EDGE)) being used by the 2.75/3G network. Performance in the GERAN network would depend on the type of FEC, feedback/ACK signal and code repetition, or lack thereof. The GERAN network supports data rates ranging from 30 kbps (code repetition, 4 time slots) to 67 kbps (6
28� HSC
GSM time slots, FEC) [14]. 15 fps QCIF video encoded using H.264 baseline profile typically requires 44.2 kbps (PSNR ~ 28 dB) [15]. 5.2 Broadcast IP Service A broadcast IP service is a unidirectional point-to-multipoint IPDC service in which data is transmitted from a single source to multiple user terminals in an associated broadcast service area. A broadcast service is a push type service, which may require subscription (similar to a multicast service), but it may not be necessary to explicitly join the service on a session basis. The typical architecture of the MBMS system is shown below:
Fig. 10 The arrows shown in 3 different colors depict three different services. For instance, cell site 1 has users who have subscribed to, and have joined the session for, all three services. Likewise, users in cell 2 are availing of only 2 services, while users in cell site 3 are availing of only 1 of the 3 possible services. In such a scenario, it is possible to split the entire service bandwidth such that each cell is serviced with only those services which the user(s) in that cell have applied for. However, since MBMS is based on the older and more traditional model of coverage planning which is a characteristic of GSM/UMTS networks, the entire coverage area of the service provider is divided into equal-sized cells, with one BTS-BSC servicing one cell. In other words, multiple carrier frequencies need to be allocated and reused across the entire coverage area for the smooth functioning of the system, thereby precluding the possibility of DVB-H-like SFN networks. The capacity of each service area cell is in direct proportion to the number of users in the coverage area of that cell. Upon a sudden increase in the number of subscribers in a cell who have joined a particular service session, the bearer services shall have to allocate smaller chunks of the total bandwidth to each user, leading to consequent degradation in the QoS for that service. Hence, the QoS limitations of an MBMS service are characterized by the following:
29� HSC
1. Low to medium data rate broadcast transmissions 2. Small to medium sized coverage areas 3. Limited scalability of service when faced with sudden escalation in the number of subscribers for
a particular service in a single service area. 5.3 Broadcast IP Service
Multicast IP based DVB-H services typically service high bit rate applications in larger coverage areas than MBMS. Moreover, the same service bits are synchronously transmitted by all transmitters within the entire coverage area of an SFN network like the one used in DVB-H. Hence, this allows some amount of scalability in the QoS in the event of a sudden increase in the number of users for a particular service. The figure below shows the feasibility of positioning a MBMS service against competing traditional 3G P2P and pure IP multicast DVB-H services. MBMS services may be suitable for low density rural/semi-urban settings with small to medium sized coverage areas. This could also include 3G femtocell scenarios, which could be used to relay 3G services (including MBMS) to a small number of users (e.g., within an apartment complex), while extending the reach of a traditional 3G network. As the user density increases, the coverage area for the MBMS service decreases in a linear fashion in order to maintain a steady QoS. Hence, their applications are severely limited in very dense urban/metro settings, which, incidentally, also include the most affluent white collar populations.
Fig. 11
6.0 DOWNLINK ANALYSIS – FRAMEWORK AND PARAMETERS This section provides a brief overview of the issues related to link budget analysis of the SC and CGC portion of a typical DVB-SH system utilizing a TR(b) gap filler transceiver. It does not cover the uplink portion of the DVB-SH system. 6.1 Assumptions Following are the assumptions made with regards to the type of satellite link used in the system architecture [20], [21]:
30� HSC
a. The earth is assumed to a perfect sphere, with a constant radius at every latitude north or south
of the equator. b. The service area of the system is the entire Contiguous United States (CONUS) comprising the
lower 48 states of the USA (66° 56‟ 58.60‟‟ W to 124° 43‟ 58.92‟‟ W, 49° 0‟ 0‟‟ N to 24° 53‟
50.52‟‟ N). c. The satellite has been placed in a Geosynchronous Earth Orbit at an altitude of (approximately)
35788.925 km AMSL. d. The inclination of the orbital plane of the satellite is 0 degrees, and the path traversed by the
satellite is assumed to be circular (eccentricity = 0).
e. The sub-satellite point is set at the median longitude of the CONUS (95° 50‟ 28.76‟‟ W, 0° N)
f. The shortest (blue) and longest (red and magenta) rays from the satellite to possible earth station locations are mapped in the diagram below. The link budget is calculated for both extremes.
g. The downlink carrier frequency for the SC has been set at 2.1 GHz (lower S Band). The CGC operates at 800 MHz (UHF).
h. The CGC employs a TR(b) class transmitter (fixed “gap filler”). i. Since the uplink is of no interest in a DVB-SH system, uplink frequencies and related link budgets
are not assumed here. 6.2 Satellite Component The basic equation used in estimating the link budget for a satellite-to-earth station downlink is a function of the transmitted power (in dBm), back-off levels and total signal path loss and system temperature. The satellite transponder radiates power across all carriers (1, 2, ….., i), which is received by all receivers (1, 2, ….., j) [20]
where,
Pmax = peak Effective Radiated Power (ERP, in dBm) of the satellite transponder channel of interest
(Pmax = Psat)
Bo,o = Back off (from saturation point) of the total output power of the satellite transponder. The total
average satellite power is now Psat = Pmax – Bo,o
AiPi = percentage of Psat assigned to the carrier of interest
G/T = this is the figure of merit for the effective receiver antenna gain (dB) against the received
system noise temperature (°K, dB) at the particular carrier frequency
G = this is the antenna gain (in dB), which accounts for the size of the receiver dish and component
losses at a particular received frequency
= Boltzmann‟s Constant (-198.6 dBm/°K-Hz)
LTj = Total losses in the receiver link (this includes free space loss, Ls, antenna pointing loss,
atmospheric and sun-outage losses, precipitation losses, radome losses, for earth station j)
31� HSC
(C/No)ij = this is the Carrier-to-Noise Power Spectral Density (psd) for the ith carrier received at
earth station j
Fig. 12 6.3 Carrier Frequency
The DVB-SH SC sub-system operates in the S band below 3 GHz. We assume an operating frequency of 2.1 GHz for the satellite link. Path Loss
The total path loss in the downlink comprises of several components, including spreading (1/r2) loss, free
space loss (Ls), ionospheric scintillations, climatic attenuation, sun outage factors etc.
Free Space Loss
The major component of atmospheric loss is Free Space Loss, Ls, which may be modeled as [20]:
32� HSC
Here, r is the total slant range from the satellite to the location of the earth station or mobile receiver.
Assuming the positioning of the satellite and earth stations for the extreme case in Fig. 12, slant range, r
may be modeled as (using the law of cosines) [21]:
where,
RE = mean radius of the earth, 6378.1 km
h = orbital height of a geostationary orbit at the equatorial plane, 35786 km
g = latitude of the earth station or location of mobile receiver; here, we assume the receiver to be
placed at the northern extremity of the CONUS (49° 0‟ 0” N)
= the difference in the longitudinal positions of the sub-satellite point (at the equator) and the earth station/mobile receiver; in our case, we assume this to be 95° 50‟ 28.76‟‟ W - 66° 56‟ 58.60‟‟ W = 28.892°
Assuming these values, the slant range is (approx.) 38855.79 km, which gives a free space loss of 190.68 dB. Attenuation due to precipitation
Rain and other forms of precipitation have a dual effect on the overall C/No figure of the link: It raises the
overall system temperature, T, and hence the G/T figure of merit of the overall link (see Antenna Figure
of Merit below), besides causing attenuation of the carrier, thereby contributing to the overall loss, Ls. It is
generally assumed that the attenuation due to all forms of precipitation is log-normally distributed (slow and flat fading). However, the attenuation due to water vapor, rain and snow is heavily frequency dependent, and is almost negligible in the S band; in fact, graphs for Mie Extinction rates (based on Laws and Parsons distribution of rain rates) [21] due not contain useful data for frequencies below 10 GHz. Thus, the overall attenuation due to all forms of precipitation in the S band may be considered to be negligible, and combined as part of the miscellaneous attenuation component of ~ 2 dB (1.52 dB due to models due to Lin and Ippolito et. al.) [21]. Oxygen and Water Vapor
Gaseous absorption is very severe in the K and higher bands, but is generally ignored (» 0.1 dB) in the S
band [21]. Ionospheric Scintillations Atmospheric scintillations in the F layer of the earth‟s ionosphere are short-term variations in the refractive index of the medium, which cause alternating signal fading and enhancements. The overall attenuation (negative or positive) may be modeled as a log-normal distribution. These scintillations are generally observed in the sub-auroral to polar and magnetic equatorial latitudes, and mostly in the night. The attenuation due to this phenomenon may be empirically modeled as [20]:
33� HSC
where,
Z = slant distance to the ionospheric irregularity in the F layer (Z lies between 400 km and 1600 km,
typical values of Z is » 600 km)
Lc = irregularity autocorrelation distance in the medium (typical value of Lc is » 1 km)
i = zenith angle at the ionospheric intersection point
= wavelength
S = Scintillation index
For short wavelength signals, like transmissions in the S band, the scintillation index may be modeled as
a function of 2; a typical value that may be assumed for a 2.1 GHz carrier is 2 dB.
Antenna Figure of Merit
The value of G/T (in dB) is a measure of the overall gain of the receiving antenna with losses due to
antenna and system temperature accounted for in the computation. The receiver antenna gain may be computed by the following empirical formula [21];
where,
GR = Gain (in dB) of the receiving antenna
= Antenna efficiency (a typical value of h is assumed to be 0.6)
Rf = carrier frequency
Da = Diameter of the receiving antenna dish
For a 2.1 GHz carrier received by a 30 m dish, the receiving antenna has a gain factor of 114 17 dB. A typical microwave receiver front-end block schematic is shown below:
The overall system temperature, Ta, as measured at the antenna input is given by the Friis‟ equation of
the general form [21]:
34� HSC
where,
T0= ambient temperature of the lossy components (290 °K)
L = Loss factor of waveguide and cable components, antenna pointing losses
TR = Effective receiver temperature (at input terminal) (assuming a cooled preamp, this is assumed
to be 100 °K)
Ta = Antenna temperature (80 °K)
F = noise figure of the modem (0.7)
Ts = system temperature of the system (assuming tandem connections)
Assuming a waveguide and cable loss factor of 4 dB (x 2.5) and a system temperature, Ts of (approx.)
114 °K, the G/T figure of merit is 21 dB/°K.
During heavy precipitation, the individual ambient and/or equivalent temperatures fall below their “clear
sky” values due to the absence of Tsun. However, due to the heavy attenuation in the line of sight signal
path, the effective system temperature, Ts may rise by about 200 °K, which is equivalent to 4.5 dB.
It is assumed that the effects of “sun-outage” is statistically improbable over a given period of time (say, 1 year), and is also mitigated by assuming an antenna with a beamwidth larger than the solid angle
subtended by the sun (1/2° squared) at the latitudes of interest.
Hence, we assume an equivalent G/T of 26 dB/°K, including margins.
Back Off The dominant source of non-linearities in the RF link is the Traveling Wave Tube Amplifier (TWTA) which acts as the backbone of the amplification sub-system in the satellite. In order to boost the signal level of the satellite transponder (so that the received signal is above the general noise floor of the RF link), the input power level of the transponder may be pushed into the saturation region of the operational characteristic of the TWTA. The amount of non-linearity is a function of the power level, and the operational point may have to pushed “back” along the locus of the characteristic curve so that any instantaneous signal amplitude variations do not lie in the “non-linear” saturation region, thereby reducing the occurrence of intermodulation components, which may increase the effective bandwidth of the transmitted signal and interfere with neighboring channels. Typically, we assume a “Back Off” factor of 3 dB to push back the operation point to the linear region of the operation characteristic of the TWTA [20], [21]. Choice of Modulation and Constellation Scheme Consider the typical sinusoidal input signal to the TWTA of the form:
The output of the TWTA is of the form:
35� HSC
where the non-linear modifiers may be parameterized using the Saleh Model [16]:
where ar, br, a and b are the parameters of the Saleh model. The amplitude and the phase of the signal
is modified as a function of the amplitude of the signal envelope, r(t). As the parameterized amplitude,
A(.) and phase, (.) are functions of r(t), the shape of a square constellation (such as 16QAM and
64QAM) is rounded at the edges and rotated along the primary axis. This results in a non-uniform spacing between individual points in the constellation. Circular constellations (such as 16APSK and 32APSK) that have a constant modulus, on the other hand, are not significantly effected by the parameterized amplitude and phase distortions [16]. Constellations which are less densely packed (such as QPSK/4QAM) are less susceptible to amplitude
variations (of the form A(.)) than higher order constellations such as 64QAM.
Hence, it is recommended to use circularly symmetric constellations (APSK family) rather than QAM constellations. Since higher data rates for DVB-SH dictate the use of higher order modulations (32APSK over QPSK or 16APSK), it is assumed that an additional “back off” of 1.5 dB is incorporated in the link budget. Co-channel interference from neighboring satellites Additional factors that may be considered in the link budget computations include:
Interference from side lobes of earth station; this is in direct proportion to the power level of the receiver antenna, and may generally be assumed to be negligible in case of mobile receivers.
Cross polarization signals from other transponders of the same satellite.
Interference from terrestrial microwave links.
Interference from adjacent satellites which may operate in the side-lobe region of the satellite‟s spectrum band.
Multicarrier intermodulation effects due to non-linearities in the TWTA. Carrier intermodulation effects require an additional power level of 17 dB to be factored into the link
budget. Likewise, adjacent satellite C/NI in the downlink is 43 dB (under clear sky conditions) and Cross
polarization losses in the downlink path require an additional 25 dB [16]. 6.4 Complementary Ground Component The terrestrial link in a DVB-SH system is a CGC component comprising a “gap filler” base station antenna and the physical media linking it to a mobile subscriber unit.
36� HSC
Gap Filler Base Station Gap filler stations may be fixed or mobile terrestrial receivers which act as relays for “forwarding” the original satellite signal to mobile and/or fixed terrestrial receivers known as Mobile Subscriber Units (MSU). Besides free space losses and interference from neighboring transmitters and channels, the biggest source of interference in this section of the link is Doppler shifts (due to relative motion between the gap filler and the mobile receiver) [16]. Downconverter The main purpose of the CGC component of the DVB-SH system is to extend the reach of the original satellite signal to users who may be out of reach of the satellite spot beam due to coverage or urban canyon issues. Since lower carrier frequencies have better penetration characteristics inside concrete, glass and steel structures, it is generally advisable to downconvert the S-band signal from the satellite to a UHF band signal. We assume an 800 MHz carrier for the CGC. An additional 4 dB loss may be assumed for the noisy components in the LNB of the amplifier in the transmitter [16]. Terrestrial Link Due to urban conditions, the reach of the terrestrial link is limited by the amount of multipath interference. Since the effects of multipath are mitigated using a Guard Interval (GI) between two consecutive OFDM symbols, the distance between the gap filler and the farthest position of the receiver (and hence the diameter of the cell) is dictated by the time taken by delayed versions of the OFDM symbol to reach the receiver antenna [16]. Antenna Figure of Merit (G/T)
Typical values of gap filler transmitter antenna figure of merit is 30 dB/°K (assuming a system
temperature, Ts of 114 °K (≡ 3 dB/°K) and including gain factors of 7 dB and 26 dB for the antenna
assembly unit and Low Noise Amplifier (LNA) block, respectively) (cf. 6.1.1.6 on how to arrive at a typical
value of receiver (GR) or transmitter (GT) gain and front-end system temperature, Ts).
The typical value of the gain of a receiver antenna in a low power hand held device ranges from (approx.) 5 dB (for omni-directional antennas) to 13 dB (for directional antennas, after accounting for a pointing loss of 2 dB). A mobile receiver is usually devoid of any cables and waveguides, and hence losses due to
these factors may be ignored. Moreover, the typical operating temperature of the devices is 290 °K (room
temperature). A typical value of G/T that may be assumed in this case is 3 dB/°K (omni-directional) to 10
dB/°K (directional).
Sun outage effects can be ignored for both the gap filler transmitter and the mobile receiver antenna, since the probability of either (or both) the antennas being within the disc of the sun is highly improbable [21], [22]. Multipath Under heavy urban conditions, interference due to several copies of the primary signal reaching the receiver (with varying levels of delay) may be modeled as a Rician process – it is assumed that there is always a line-of-sight from the gap filler to the receiver antenna, besides the reflected versions of the signal.
37� HSC
A typical level of degradation that may be assumed in such a situation is (approx.) 25 dB [16]. Receiver Sensitivity and Maximum Detection Range
Receiver Sensitivity, Smin of a black box receiver is a measure of the minimum input signal needed to
produce a specified output signal, given a fixed signal-to-noise (S/N) ratio, and is defined as the
minimum Signal to Noise Ratio (S/N)min multiplied by the mean noise power.
This measure is especially useful in determining the maximum reach of a transmitted signal, given a certain SNR performance characteristic of the receiver[21].
where,
Rmax = maximum range of the transmitted signal, assuming a minimum SNR (S/N)min at the
receiver
Pt = Power level of signal at the transmitter antenna
Gt = Gain of transmitter antenna
Gr = Gain of receiver antenna
= wavelength of transmitted signal (0.375 m for a UHF signal)
T0 = absolute system temperature of the receiver block (assumed to be 290°K)
B = system bandwidth of the receiver block (assumed to be 8 MHz)
NF = Noise Factor of the receiver
We assume a transmitter power level, Pt of 30 W at the gap filler transmitter antenna front end, a
transmitter gain, Gt of 20 dB (after accounting for a 10 dB drop due to system and atmospheric noise and
cable and other losses), a receiver gain, Gr of 5 dB (for an omni-directional antenna with negligible
pointing losses) and a noise factor, NF of 6 dB (for a solid state receiver in the MSU).
The minimum SNR of the receiver system, Smin is assumed to be 16 dB. This value is dependent on the
type of application as well as the nature of signal detection (auto or manual/human detection). In the case of a mobile reception scenario, not only does the amplitude need to above the noise floor, but the carrier frequency also needs to be maintained within tolerance levels.
This gives a maximum reach, Rmax of the signal, and hence the maximum cell size of 4.53 km.
Propagation Loss The average size of each cell is computed based on the duration of the useful OFDM symbol and the GI. The radius of the cell is given by the empirical expression:
38� HSC
where,
rCELL = radius of the cell; this is also the furthest distance traveled by a signal from the gap filler
base station
TU = useful time period of OFDM symbol time
Tg = guard interval of OFDM symbol time
= GI as a factor of total symbol time
c = speed of light in free space = 3 x 108 m/s
I have assumed the semi-deterministic COST231-Walfisch-Ikegami Model (COST231-WI) (cf. [19], [20] for details) in order to model the propagation characteristics of the terrestrial channel in an urban-canyon environment. The model may be empirically represented as follows:
where,
L0 = free space loss
Lrts = roof top to street diffraction loss
Lmsd = Multiscreen diffraction loss
For further details on the empirical formulation of each of the three components in the COST231-WI model, cf. [18] and Appendix A of [17]. Assuming an 800 MHz UHF carrier in the CGC link, a cell radius of 4 km, a base station antenna height of 30 m, average roof height of 20 m, mobile receiver height of 1.8 m, a building spacing of 50 m, street
width of 30 m and a street orientation of 90[17], the typical loss in the link is (approx.) 126 dB.
Atmospheric Attenuation UHF band signals are less susceptible to attenuation to different types of precipitation such as rain and snow. However, the effective antenna noise temperature (under heavy precipitation conditions such) may be assumed to be typically about 100 K above the clear sky value, which introduces a 1.5 dB margin in the effective noise level of the terrestrial link. Since much of the useful carrier power does not have to pass through the F layer of the ionosphere, losses due to scintillations may be ignored along the terrestrial downlink [21].
7.0 FADING CHARACTERISTICS OF DVB-H AND SC AND CGC FOR DVB-SH 7.1 Rician fading
The DVB-H transmission system as well as the SC of DVB-SH suffers from Rician fading. Under certain special conditions (slow moving receivers, for e.g.), this model may also be applied to TR(a) based CGC components.
39� HSC
This fading model is caused by Doppler shifted echoes with a Normal Probability Distribution Function (pdf). In addition to the multipath component, there is always a direct path from the transmitter antenna to the receiver module, known as the Line of Sight (LoS) [26]. The LoS component would typically form the major component in the fading model, and considerably boosts up the receiver field strength. Several Doppler shifted echoes which are received at the Rx antenna would arrive from different
directions, and hence the fading model accounts for all values of from 0 to 360°. The angle with the
lowest probability is at = 90°, since the vehicle will only in rare cases be moving around the Tx antenna
in circles. Consequently, the maximum fading (and lowest signal power level) would be found at the carrier center frequency, f. The highest receiver signal power levels, and hence the least amount of fading is observed at frequencies
f fD, since the probability of the receiver moving directly towards ( = 0°) or away from ( = 180°) the
transmitter antenna is the highest. Between 0 and 180, the distribution of the fading function is in the form of a bell shaped (Gaussian) curve. This is produced by the Doppler shifted echoes produced at random by reflections from obstacles that keep coming in (and going out of) the path of the mobile receiver. A similar normal pdf may be obtained if we measure the received signal strength when the receiver antenna is moving around the transmitter and changing the direction at random. The spectral bandwidth results from the maximum Doppler shift of the single echoes, and the Rice peak results from the Doppler shift in the LoS path. These two components are almost never correlated, since the probability of the angles of the receiving path (of the single path echoes and the LoS) coinciding is extremely munute. 7.2 Rayleigh fading
The CGC of the DVB-SH (in particular, the TR(b) and TR(c) type transmitters) may be modeled using Rayleigh fading models. Rayleigh fading is modeled in a fashion similar to Rician fading, although the direct LoS path is missing in the former case. The received power as a function of time exhibits attenuation produced by the superposition of all the echoes with different power levels and phases at the Rx antenna. The maximum attenuation due to Rayleigh fading may reach upto 60 dB because the continuous power component of the LoS path is missing [26]. 7.3 Log normal (SUZUki) fading Both Rayleigh and Rician fading are considered “fast” profiles, and track changes in the received power levels over distances as short as the wavelength of the received signals; in the case of DVB-H (UHF) and DVB-SH (S Band) signals, this would range from 142 to 375 cms (typically, < 0.5 m). In most real world scenarios, such distances would be covered in a matter of a few milliseconds. Log-normal (or, Suzuki) fading models, on the other hand, track “slow” changing profiles in open or flat terrains, where the characteristics of the environment do not change over large distances (in the range of several tens of meters) [26]. Log-Normal fading always accounts for Rayleigh type Doppler shifted echoes, and only occasions have to factor in a LoS component like in Rician models. This modeling profile is suitable for rural settings, and hence is suitable while modeling the SC of DVB-SH.
40� HSC
8.0 PHY/MAC LAYER PERFORMANCE It is generally assumed that certain station keeping maneuvers are required to maintain the inclination of
the satellite orbit at precisely 0° w.r.t the equatorial orbit. However, due to the gravitational force applied
by the sun and the moon on the satellite, a drift of approx. 0.85°/year [21] is observed, which translates to
a maximum inclination of (approx.) 14.67° in 26.6 years, and about ± 3° [21] in a 10 year mean life span
of a typical geosynchronous satellite. As shown in 4.1, one of the major effects of an inclined orbit of a GEO satellite is the variation in the instantaneous distance from the satellite to the sub-satellite point, due to Kepler‟s Laws. This also
translates into an instantaneous variation in the slant height, H from the satellite to any point on the
surface of the earth other than the sub-satellite point (for e.g., the location of the earth station). As a result of this orbital eccentricity, two different phenomena are observed, which affect the two operational modes (-A and -B) of the SC of the DVB-SH system in different ways. 8.1 Doppler effects in DVB-SH-A During the design of the PHY layer sub-systems, the sub-carriers in the OFDM symbol transmitted by the DVB-SH-A SC carrier are preset to extremely precise values, which is helpful in letting the receiver block to lock on to the carrier with minimal assistance from pilot information. For e.g., a system operating in a 6 MHz channel and utilizing the 4K mode (4096 sub-carriers) would require the sub-carriers to be at a separation of precisely 1674.107 Hz [7] from one another. The system is typically designed for an average value of instantaneous velocity of the satellite, and any deviation from the average (e.g., at the apogee or perigee) would translate to a positive or negative Doppler shift in the carrier frequency of the transmitted signal. For instance, a 10 Hz to 75 Hz (cf. 4.1.1 for details) carrier shift means a ± 0.5% to 4.5% deviation from the average. If the deviation is beyond a particular threshold (along the frequency axis), the sampled carrier amplitude may fall below the noise floor, and hence may be completely ignored by the receiver, thereby leading to errors. Hence, in order to mitigate the effects of frequency shifts, it is recommended to use a broader pulse shape (such as raised cosine, instead of a sinc waveform) while generating the OFDM sub-carriers, so that the pulse is above a minimal threshold for a wider ranger of neighboring frequencies, and any deviation in the position of the carrier would still be recorder above the noise floor. 8.2 Variation in downlink trip delays in DVB-SH-B The SC of the DVB-SH-B operational mode uses a TDM carrier waveform, which allocates fixed time slots for transmitting payload data to specific users tuned to the same carrier/channel. The start time and end time of each time slot (allocated to a particular user), the duration of the slot, and the time gap till the next time slot (for the same user) have very precise values with extremely minute tolerance limits.
Due to a slightly elliptic inclined orbit [16], the slant distance, H from the satellite to the earth station varies
from the average value at the apogee and perigee locations. This causes a delay (positive for apogee, and negative for perigee) in the transmitted signal from reaching the earth station. In order to accommodate the delay, a “Guard Interval” has to be appended after each time slot so that a “delayed” signal may not spill over to a neighboring time slot meant for another user. This reduces the number of time slots that may be fitted into a given time period, thereby reducing the number of users that may be supported by the channel.
41� HSC
Changes in the inclination of the orbital plane of the GEO satellite causes the apogee and perigee distances to vary at a rate of (approx.) 350 km/year, resulting in a recalibration in the timing graph of the
TDM system of about ± 0.1% every year of the lifetime of the satellite.
8.3 Choice of modulation scheme in the mapper Cf. section 6.1.1.8 for a detailed discussion on the choice of APSK class constellation maps over QAM/QPSK class constellations for the SC. 8.4 Performance characteristics of MPE-FEC time interleavers
Time interleaving is performed in both PHY and MAC layers of the DVB-H and DVB-SH systems. The IP data from the transport/network layer is fed into the MPE-FEC matrix in a column wise fashion. After FEC coding (using RS64), the output is fed into corresponding “sections” (similar to datagrams) in a row wise fashion. Thus two consecutive bytes in the IP datacasting stream are now separated by a distance of n bytes, where n is the number of rows in the MPE-FEC matrix (and has fixed values of 256, 512, 768 or 1024). In mobile environments, the signal may be subjected to short term
fading. The frequency components in the carrier signal which are affected by this fading depends on the amount of Doppler the receiver is experiencing, which in turn is a direct function of the relative velocity between the receiver and the transmitter. As a general rule, lower the Doppler shift, greater is the time interleaving required in order to offset the affects of fading. As a general rule, a larger sized MPE matrix is required to handle higher relative velocities between the Tx and Rx modules. Figure 13 [2] demonstrates the C/N performance of the carrier signal for scenarios which employ time interleaving using MPE-FEC as well as situations which do not use this feature. Performance is a direct function of various PHY and MAC layer parameters such as FFT mode, GI interval, modulation scheme and MPE-FEC coding rate. It is clear from the figure that MPE-FEC allows a particular level of QoS to be achieved at a C/N level which is approx. 5 dB lower than the situation when MPE-FEC is not implemented (upto Doppler shifts of 100 Hz). For extremely high speeds (Doppler shifts of greater than 100 Hz), performance converges at about 120 Hz, and hence MPE-FEC and any additional time interleaving has no tangible benefits. Moreover, for Doppler shifts beyond 120 Hz, the sub carriers tend to overlap for all modes (1K for DVB-SH, 2K, 4K and 8K for DVB-H), and hence such velocities may not be supported. 8.5 Performance characteristics of native and in-depth interleavers Various studies in the open literature ([3],[23]) have studied and demonstrated the efficacy of using both native as well as in-depth interleavers in the PHY layer of DVB-H and DVB-SH systems. In the case of DVB-H, of special interest is the performance enhancement when the 4K FFT mode is used in conjunction with in-depth interleavers [23].
42� HSC
In most simulations, the performance of the system is measured w.r.t Bit Error Rate (BER), the number of bits in the bit stream received in error after all levels of error correction (Reed Solomon, Convolutional, Turbo (in the case of DVB-SH)) have been executed. Under fixed reception conditions, a standard Signal-to-Interference Ratio (SIR) of -5 dB is set; this corresponds to a impulse noise source (caused by a radiating electrical device, or the ignition system of a car, for example) whose power level is set at 5 dB below the signal power level. A typical set of PHY and/or MAC parameters used include 16-QAM modulation, Convolutional code rate of 2/3 and a GI of ¼ [23].
As shown in Fig 14 [23], the performance of the link is a direct function of the FFT mode and the type of native or in-depth symbol interleaver. In the case of in-depth interleaving, only the 2K and 4K modes of DVB-H are studied, while both the native as well as the more memory efficient in-depth versions are compared for the 8K mode. The SIR is measured and set at a fixed value by overlaying a known transmitted signal with a fixed pattern of disturbances maintained at a fixed power level. It is generally observed that in-depth interleaving in the 4K and 2K modes outperform the native interleavers under similar conditions. However, the native interleaver under the 8K mode is generally considered
the most ideal candidate for ameliorating the effects of both short-term fading and multipath. However, the downside of using native interleavers is that the BER performance suffers for a given Doppler frequency shift, as the FFT size increases, thereby offsetting any gains from using a higher FFT mode. This also restricts the maximum relative velocity between the Tx and Rx modules that may be supported for a given data rate and QoS. 8.6 Tradeoffs in time-interleaver design in the MAC layer As discussed earlier in this document, the DVB-SH system specifies 2 types of receivers, based on the type of time interleaving and the length of the interleaver. Class 1 receivers interleave a single PHY layer block, which are able to mitigate short-term fading and occlusion/blocking by obstacles. These phenomenon are temporary in the case of a fast mobbing receiver, and do not affect the overall QoS of the system. Moreover, the memory size, and hence receiver cost is a major factor in considering this type of receiver specification. Additional protection is performed by the MPE-FEC in the LLC layer.
43� HSC
Class 2 receivers perform PHY layer interleaving across several OFDM symbols, which mitigate long term fading effects which may or may not be caused by the relative motion between the Tx and Rx modules. If allowed to persist, long term fading (upto 10 seconds) could severely hamper the QoS rating of the system, and hence are more critical to the overall system performance. However, larger on-chip memory and processing power is required of the processing platform, and this has bearing on the overall cost of the system. Additionally, depending on whether PHY or LLC layer FEC is employed, adaptation of the waveform to the receiver constraints is more progressive [25].
9.0 VIDEO AND AUDIO PROFILES SUPPORTED BY DVB-H AND DVB-SH The DVB-H standard specifies a maximum bit rate of 31.7 Mbps [4], [5], [24] for any FFT mode and modulation pattern. If the punctured inner code rate (Convolution for DVB-H) is represented as CR, the outer code rate (Reed Solomon 16 for DVB-H and Turbo for DVB-SH) is RSR, the Guard Interval is GI, the bits carried per sub carrier of the OFDM symbol is M, the number of active carriers in a symbol is N and the symbol duration is T, then the bit rate is defined as:
The upper limit of the supported bit rate is achievable by setting CR = 7/8, GI = 1/32, M = 6 for 64 QAM. N could be 853, 1704, 3409 or 6817 for 1K, 2K, 4K or 8K FFT modes, while T can take a value of 179.2,
358.4, 716.8 or 1433.6 s respectively. Inserting these values into the equation above, we observe that the maximum bit rate ranges from 11 Mbps to 22 Mbps. The table below displays the various H.264/AVC profiles and levels, as specified by the ISO/IEC IS:14496-10 specification [25].
Profile Level Footprint Max. bit rate
Simple L0, L1 QCIF 64 kbps
L2 CIF 128 kbps
L3 CIF 384 kbps
Advanced Simple L0, L1 QCIF 128 kbps
L2 CIF 384 kbps
L3 CIF 768 kbps
L4 VGA 3 Mbps
L5 VGA 8 Mbps
Advanced Real Time Simple L1 QCIF 64 kbps
L2 CIF 128 kbps
L3 CIF 384 kbps
L4 CIF 2 Mbps
Profiles specify a specific tool set used by the codec to achieve the required encoded bit rate performance. Levels specify the physical foot print of the visual area of the video stream. As is seen from the data in the table, multiple combinations of video streams may be transmitted over time sliced DVB-H or DVB-SH transmission systems based on the video footprint and encoding profile. For e.g., about 4 to 10 different ARTS-L4 (for handheld PDAs) video streams may be transmitted within the 11 Mbps to 22
44� HSC
Mbps bit rate range computed above (about 20% overhead in redundancy coding in the MAC, Network and Codec layer is assumed). Similarly, a maximum of 2 AS-L5 (for laptops) video streams may be transmitted within the same bit rate range.
10.0 REFERENCES 1. Mobile Broadcast/Multicast Service (MBMS): White Paper, Aug 2004, Media Lab, TeliaSonera,
Finland
2. DVB-H: Digital Broadcast Services to Handheld Devices, Gerard Faria, Jukka Henriksson and Erik Stare
3. On the use of OFDM for beyond 3G Satellite Digital Multimedia Broadcasting, Cioni, Corazza, Neri and Vanelli-Coralli
4. Draft ETSI TS 102 585, European Telecommunications Standards Institute
5. Draft ETSI EN 302 583, European Telecommunications Standards Institute
6. ETSI EN 300 468 v1.7.1, European Telecommunications Standards Institute
7. ETSI EN 300 744 v1.5.1, European Telecommunications Standards Institute
8. ETSI EN 301 192 v1.5.1, European Telecommunications Standards Institute
9. ETSI EN 302 304 v1.1.1, European Telecommunications Standards Institute
10. ETSI TS 101 191 v1.4.1, European Telecommunications Standards Institute
11. ISO/IEC IS:13818 MPEG Part 2, International Standards Organization/International Electrotechnical Commission
12. Design and Performance of Predistortion Techniques in Ka-band Satellite Networks, Salmi, Neri and Corazza
13. Minimizing the Peak-to-Average Power Ratio of OFDM Signals Using Convex Optimization, Agarwal and Meng
14. Provision of MBMS over the GERAN: Technical Solutions and Performance, Provvedi, Rattray, Hofmann and Parolari
15. Improved H.264/AVC video broadcast/multicast, Tian, Kumar MV, Hannuksela, Wenger and Gabbouj
16. Physical Layer Impairments in DVB-S2 Receivers, Nemer
17. Channel Models for Fixed Wireless Applications, Erceg, Hari, et. al.
18. Propagation Prediction Models, Cichon and Kürner
19. A theoretical Model of UHF Propagation in Urban Environments, Walfisch and Bertoni
20. Propagation Factors Controlling Mean Field Strength in Urban Streets, Ikegami, Yoshida, Takeuchi and Umehira
21. Digital Communications by Satellite, James Spilker, ed. Thomas Kailath
22. Satellite Communications Engineering, Pritchard, Suyderhoud and Nelson
23. DVB-H – The emerging standard for mobile data communication, Kornfeld and Reimers
24. On typical service bit rates and delivery in DVB-H networks, Jashek and Meiri
25. H.264 and MPEG-4 Video Compression: Video Coding for Next Generation Multimedia
26. An Extended Suzuki Model for Land Mobile Satellite Channels and Its Statistical Properties, Patzold, Killat and Laue