channel element dimensioning guideline

ERICSSON WCDMA RADIO ACCESS NETWORK

CHANNEL ELEMENTDIMENSIONING GUIDELINE

62/100 56-HSD 101 02/5 Rev A 2007-03-20 ERICSSON CONFIDENTIAL INFORMATION 1(29)

Ericsson AB 2007

The contents of this product are subject to revision without notice due to continued progress in methodology, design and manufacturing.

The document has obtained the classification ERICSSON CONFIDENTIAL. This means that it can not be delivered to customers or be used for marketing purposes or be distributed internally outside the approved distribution channels, as determined by FJW/SM management.

CHANNEL ELEMENT DIMENSIONING GUIDELINE

Revision history

Rev Date Description

A 2007-03-26 Document replacing 62/1551-HSD 101 02/1 Rev C.

2(29) ERICSSON CONFIDENTIAL INFORMATION 62/100 56-HSD 101 02/5 Rev A 2007-03-20


Contents

1 Introduction........................................................................51.1 Purpose............................................................................................5

1.2 Overview...........................................................................................5

1.3 Abbreviations....................................................................................5

1.4 Limitations.........................................................................................6

2 Definitions and assumptions...........................................62.1 Definitions.........................................................................................6

2.2 Assumptions.....................................................................................8

3 Channel Element resource model....................................93.1 Number of users per site...................................................................9

3.2 Channel element cost model.............................................................9

3.3 Basic resource model......................................................................10

3.4 Resource model for Enhanced Uplink..............................................10

4 Dimensioning method.....................................................114.1 Overview.........................................................................................11

4.2 Identifying the average number of simultaneous users.....................12

4.3 Identifying the peak number of simultaneous users..........................15

4.4 Calculating number of CEs..............................................................15

4.5 Special traffic cases........................................................................19

5 Examples..........................................................................205.1 R99, only conversational traffic admitted in busy hour......................20

5.2 R99, conversational and best effort traffic admitted in busy hour......22

5.3 R99 + HSPA, only conversational traffic admitted in busy hour........24

6 CE Observables...............................................................266.1 Introduction.....................................................................................26

6.2 CE utilization using observables......................................................26

6.3 Estimation of KSHO,UL and KSHO,UL.......................................................27

7 References........................................................................30



1 Introduction

1.1 Purpose

The purpose of this document is to present methods to dimension the number of channel elements required per site. The document assumes that a traffic mix has been specified for the WCDMA Network to be dimensioned, broken down to site level. The dimensioning methodology is based on the functionality supported by WCDMA RAN P5. The document also presents a method for dimensioning or re-dimensioning in a commercial network, when the channel element utilization has been analyzed by statistics counters or other observables.

The main target groups for this document are radio network design engineers and technical sales support engineers.

1.2 Overview

This document is outlined in the following manner: Chapter 2 presents definitions and assumptions that are relevant for channel element dimensioning, such as impact of channel switching functionality, HSDPA and Enhanced Uplink (EUL). Chapter 3 presents a channel element resource model. This chapter describes the general model of channel element cost and how to calculate channel elements when the exact number of users with different services per site is known. Chapter 4 contains the dimensioning method to use when the only input is offered traffic. It can be used for pure R99 traffic as well as for R99 traffic mixed with HSDPA and EUL. Chapter 5 gives a number of examples of calculating the number of channel elements using the method in chapter 4. Chapter 6 presents a method how to dimension channel elements based on statistics and other observables.

1.3 AbbreviationsA-DCH Associated Dedicated Channel

AF Activity Factor

AMR Adaptive Multi Rate

BE Best Effort

bps Bits Per Second

CE Channel Element

CS Circuit Switched

DCH Dedicated CHannel

ddm Discrete Distributed Measurements

DL DownLink

E-DCH Enhanced Dedicated Channel



EUL Enhanced Uplink

GoS Grade of Service

HS High Speed

HSDPA High Speed Downlink Packet Access

HS-DSCH High Speed Downlink Shared Channel

HSPA High Speed Packet Access

HW HardWare

MO Managed Object

PS Packet Switched

R99 Release 99

RAB Radio Access Bearer

RAN Radio Access Network

RB Radio Bearer

RBS Radio Base Station

SHO Soft Handover

SW Software

UE User Equipment

UL UpLink

WCDMA Wideband Code Division Multiple Access

1.4 Limitations The document is valid for WCDMA RAN P5.

This document does not describe RBS HW dimensioning.

This document does not consider any settings of the maximum number of HSDPA and EUL users per cell.

This document assumes that a certain traffic mix has been specified. No description of traffic models are given in this document.

2 Definitions and assumptions

2.1 Definitions

2.1.1 Channel Element

Channel Element (CE) describes the SW licensed capacity resources required for a dedicated channel (DCH or E-DCH). The number of channel elements, nCE, required in a RBS is based on the traffic type, and is dependent on the radio



bearers to be used, as well as on the number of simultaneous users for each specific radio bearer.

2.1.2 Hardware

Each time a dedicated channel is allocated, channel elements will be consumed in UL and DL. The consumption can be described as a cost, and depends on the type of radio bearer for the dedicated channel. The channel element cost can be expressed with weighting factors for the various radio bearers, the so called channel element factors, DL and UL.

If there is an insufficient amount of channel elements available, blocking might occur. If possible, to avoid blocking, a best effort user will be switched down to a lower rate to make channel elements available for the requested dedicated channel.

The channel element capacity depends on the HW installed in the RBS as well as on the channel element SW licensing restrictions. To obtain the maximum HW capacity for different RBS equipment, see the RBS 3000 Baseband Product Description [1].

2.1.3 HSDPA

Traffic on the HS-DSCH will not consume channel elements. However, every HSDPA connection will require A-DCHs for UL/DL signaling and UL traffic. A-DCHs are dedicated channels, which consume channel elements. Dimensioning of DL resources for A-DCH is not needed since DL A-DCHs do not consume resources from the normal pool of SW licensed channel elements. However, resources for UL A-DCH have to be considered since they consume channels elements from the normal SW licensed pool.

2.1.4 Enhanced Uplink (EUL)

EUL can only be used for connections that use HS-DSCH in the DL. Each admitted EUL user will carry traffic and signaling on an E-DCH. Similar to HSDPA, the E-DCH user has one serving cell. The E-DCH serving cell is always equal to the HS-DSCH serving cell for the connection. If the terminal is in soft or softer handover there is always at least one non-serving E-DCH cell.

E-DCH users consume channel elements from the SW licensed CE pool, but the resources for E-DCH are handled in a different way compared to A-DCH and DCH. The resources for E-DCH are controlled by the EUL scheduler located in the RBS and resources for A-DCH and DCH are controlled by the RNC.

Each admitted E-DCH user will have a minimum amount of channel elements allocated. The minimum amount of channel elements depends on licensing; see [1].Those channel elements are reserved to the E-DCH user as long as the user is active in the cell. The minimum channel element allocation can be used for non-scheduled data (signaling) and scheduled data to a certain rate (user data). The channel elements represented by the minimum CE allocation are guaranteed and are not affected by Enhanced Soft Congestion or re-scheduling in the RBS. As a



result, an admitted E-DCH user has the possibility to reach a minimum rate supported by the minimum CE allocation.

If an E-DCH user requests rates higher than the rates supported by the minimum CE allocation, additional channel elements will be needed. These are handled by the EUL scheduler only and are not seen by the Admission Control function. The EUL scheduler will grant an E-DCH user a higher rate if it is possible to allocate more CEs in addition to the other two required UL resource types (air interface interference and Iub bandwidth [2]). The additional CEs for an E-DCH user might be affected by Enhanced Soft Congestion or re-scheduling in the RBS.

If the minimum amount of channel elements is not available at the E-DCH establishment attempt it will not be possible to admit a new E-DCH user.

2.2 Assumptions

2.2.1 Channel rate switching

Channel rate switching, which is a part of the channel switching functionality, is an important aspect of packet traffic from an equipment dimensioning point of view. It should be recognized that channel rate switching can change the radio bearer data rate from 64 kbps up to what the system is capable of, which can be 128, 384 kbps or EUL/HS. In this document, channel switching up to EUL/HS is assumed when the interactive class traffic is considered. The channel rate switching between two dedicated channels is assumed to be ideal.

Channel rate switching can be triggered by throughput, coverage or soft congestion criteria.

2.2.2 Enhanced Soft Congestion

This document contains examples with and without the impact of Enhanced Soft Congestion. Enhanced Soft Congestion control is a function that aims at increasing accessibility. It allows a user to be admitted in the cell when the admission limits are exceeded and another best effort user can be switched down in rate. When admitting a user the system evaluates whether or not a downswitch of an existing best effort user shall be initiated. Only best effort users on dedicated channels are target for enhanced soft congestion and a user can never be downswitched to a lower rate than 64/64.

The channel rate switching due to Enhanced Soft Congestion is assumed to be ideal.

2.2.3 Impact of common channels

It is not necessary to dimension for common channels, since common channels do not consume channel elements.



2.2.4 Compressed mode

Allocation of channel elements due to compressed mode has not been considered, since there is no additional channel element cost for connections in compressed mode.

2.2.5 Trunking per RBS site

It is worth noting here that the offered traffic as an input to CE dimensioning is per RBS cabinet including all carriers and all sectors. The RBS can be seen as having a pool of hardware resources, which enables trunking of connections.

3 Channel Element resource model

3.1 Number of users per site

If the number of users (simultaneously connected) per site is known the following expression can be applied to calculate the number of channel elements needed:

(1)

where

Ni is the number of currently connected users per site for service i.

i is the channel element factor for service i.

3.2 Channel element cost model

Table 1 gives an example of channel element factors, showing the channel element consumption for different services on different radio bearers. Please note that this is an example and the cost-model may differ depending on RBS and CE licensing. To obtain the actual figures for CE factors and CE capacity, please refer to RBS 3000 Baseband Product Description [1].

Table 1. Example of channel element factors

Service CE factor

UL DL

AMR12.2 1 1

PS64 4 2

PS128 8 4

PS384 16 8



3.3 Basic resource model

Equation 1 and Table 1 shows that the number of channel elements required per site depends on number of simultaneously connected users and service types for the connected users. This basic resource model will be valid for the dimensioning methods to be described in this document. However, some adaptation will be done, since the methods to be described are based on offered traffic and not on a known number of connected users.

3.4 Resource model for Enhanced Uplink

The resource model for Enhanced Uplink differs from the basic resource model, since resources for E-DCH users are handled by the EUL scheduler.

As described in section 2.1.4 an admitted E-DCH user will have a minimum amount of CE reserved for signaling and scheduled user data up to a certain rate. If the E-DCH user request higher rates additional CEs are needed. The user will be scheduled higher rates if additional CEs and other UL resources are available. The additional CEs for higher rates are not guaranteed and might be target for re-scheduling (if other E-DCH user request higher rates) or Enhanced Soft Congestion. This can be seen in Figure 1.

Figure 1. Channel element allocation for EUL

Figure 1 shows that if the R99 traffic load is low in the RBS the EUL scheduler will have more channel elements to share between the EUL users. The EUL users will thus have a higher probability of being scheduled to high rates at low R99 traffic. If R99 traffic load increases, channel elements will be removed from the EUL domain and EUL users will be scheduled to lower rates.

The maximum number of channel elements that can be allocated by E-DCHs can not be higher than the EUL license, even if there are channel elements available in the SW licensed pool. Table 2 gives an example of channel element factors for different E-DCH user data rates. Similar to the basic resource model, the channel element cost for an E-DCH user increases if the E-DCH user rate increase. Please note that this is an example and the cost-model may differ depending on RBS and


EUL license

SW licensed CE pool

HW dependent

CE capability

HW dependent

CE capability

DCH

Initial minimum CE allocation (per EUL

user

E-DCH


CE licensing. To obtain the actual figures for CE factors and CE capacity, please refer to RBS 3000 Baseband Product Description [1].

Table 2. Example of channel element factors for EUL

SF CE factor (UL) Bit rate (kbps)

256 2 1.8

128 2 –

64 2 –

32 2 19 – 37

16 4 37 – 71

8 8 71 – 155

4 16 155 – 711

2x4 32 711 – 1448

4 Dimensioning method

4.1 Overview

The dimensioning method consists of two main parts:

Identifying the number of simultaneous users – average and peak number of simultaneous users is described in section 4.2 and 4.3.

Calculating the number of CEs – average and peak traffic is considered and described in section 4.4.



Figure 2. Channel elements dimensioning method

4.2 Identifying the average number of simultaneous users

4.2.1 General concept

According to equation 1, the number of channel elements can be calculated if the number of users is known for each service. It is a little more complex when all that is given is offered traffic per site. In this case the mean holding time for the individual radio bearer has to be considered.

In cases where application specific characteristics are not given by the customer, it is valid to simplify the dimensioning process by assuming that all traffic carried on an individual radio bearer has the same characteristics. An approach to calculate the number of simultaneous users can then be applied by specifying an average user. The following calculations use that assumption.

In case the offered traffic Ai is given in bps, the following expression can be used to convert to offered traffic in Erlang:

(2)

where

A i,bps is the offered traffic in bps for service i.

Ri is the bearer rate for service i.

KAF, i is the activity factor for service i.

The activity factor must be included since it models the difference between the minimum time needed for transmission with the peak rate and the session time.


6. Finding the total number of CEs required

3.Calculating nCE,Peak

4.Calculating nCE,AVE

5.Calculating nCE,BE

Calculating the number of CEs

Identifying the number of simultaneous users

2. Identifying the peak number of simultaneous users

1. Identifying the average number of simultaneous users


The activity factor will allow more simultaneous users on the air interface and therefore more users allocated on a dedicated channel.

In many data applications the offered traffic is asymmetric with respect to UL and DL. For a ratio of 1:10 of UL and DL offered traffic is common. When a user is set up on a DCH, the resources in UL and DL will be occupied, although one of the links will be very under-utilized. Thus, in the calculations, the offered traffic for hardware dimensioning is defined as follows:

(3)

where

AUL,i is the offered UL traffic per site for service i given in Erlang.

ADL,i is the offered DL traffic per site for service i given in Erlang.

It is worth noting here that the offered traffic A is given per RBS, including all carriers and all sectors. The RBS can be seen as having a pool of hardware resources, which enables trunking of connections. It should also be noted that retransmissions are included in the traffic.

The offered traffic must also take soft handovers into consideration. The offered traffic for a certain service including soft handover is given by the following expression:

(4)

where i is the fraction of soft handover users for service i.

4.2.2 HSDPA considerations

CE dimensioning for DL HSDPA related traffic does not need to be considered, since CEs for HS-DSCH and DL A-DCH are not taken from the normal pool of channel elements. However, UL A-DCH CE resources should be dimensioned carefully, to prevent situations where a user cannot be scheduled data on DL HS-DSCH due to lack of CE in the UL although there are enough resources in the DL.

CE dimensioning can be based on offered UL A-DCH traffic. If offered UL A-DCH traffic is not specified, then the UL CE dimensioning can be based on HSDPA DL traffic, since UL A-DCH will be allocated as long as the user might be scheduled data on HS-DSCH. This means that the UL A-DCH holding time is the same as the DL HS-DSCH holding time.

If EUL is supported on the uplink the distribution of DCH/HS and EUL/HS traffic has to be taken into account. Also the 64/HS and 384/HS traffic distribution has to take into account. EUL and A-DCH traffic distribution can be obtained from a traffic model or from analysis of counters or other observables.

4.2.3 EUL considerations

EUL traffic is regarded as best effort traffic, since it makes use of resources remaining after the guaranteed resources have been served. However, EUL CE



resources should be dimensioned carefully to support an EUL target rate in the cells, soft/softer handover for E-DCH users and to prevent situations where users cannot be scheduled data on DL HS-DSCH due to lack of CE in the UL for EUL.

CE dimensioning can be based on offered EUL traffic. If EUL traffic is not specified, the UL CE dimensioning can be based on HSDPA DL traffic, since E-DCH will be allocated as long as the user might be scheduled data on HS-DSCH. This means that the E-DCH holding time is the same as the DL HS-DSCH holding time.

Since HSDPA related UL traffic might also be carried on UL A-DCH, the distribution of DCH/HS and EUL/HS traffic has to be taken into account. EUL and A-DCH distribution can be obtained from a traffic model or from analysis of counters or other observables.

EUL in soft handover

The soft handover traffic for EUL is handled in a different way compared to R99 and DCH/HS traffic. The reason is that the EUL scheduler applies dynamic CE allocation for serving cell E-DCH users and fixed CE allocation for non-serving cell E-DCH users.

The EUL scheduler attempts to distribute resources so that each serving cell E-DCH user ends up with at least a rate defined by eulTargetRate. In loaded traffic situations Enhanced Soft Congestion might trigger the EUL scheduler to schedule serving cell E-DCH users to rates below eulTargetRate. However, the rate for a serving cell E-DCH user will never be lower than the rate supported by the minimum CE allocation, see section 3.4.

The fixed CE allocation for non-serving cell E-DCH users in soft handover is defined by eulReservedHwBandwidthSchedDataNonServCell. The fixed CEs for those users are not targeted in the function Enhanced Soft Congestion.

CE factors corresponding to eulTargetRate, minimum HW allocation and eulReservedHwBandwidthSchedDataNonServCell are licensing dependent and based on the resource model for Enhanced Uplink, see section 3.4.

4.2.4 Statistical multiplexing/complete partitioning

With only one service, it would suffice to use an Erlang B table to find the maximum number of users for a given grade of service. In case of several services, this method can be used if the resources are separated and each partition is dedicated to a single resource. This is called complete partitioning and gives an overestimation of the required resources.

A more accurate but also more complicated method is to apply statistical multiplexing. This method utilizes the fact that the probability for all services having their peak at the same time is very low. This approach results in lower total load than the complete partitioning. The difference can be expressed as a statistical gain. This gain is illustrated in Figure 3. To be able to use the statistical multiplexing approach, a calculation tool is required since the computations are iterative and can get complicated. Statistical multiplexing will not be further described in this document.



A theoretical discussion on statistical multiplexing plus practical examples is given in the document Statistical Multiplexing, Theory and Implementation [3].

Figure 3. Comparison between complete partitioning and statistical multiplexing

4.3 Identifying the peak number of simultaneous users

When the only input for channel element estimation is the offered traffic, the estimation has to take peak traffic into consideration. By assuming a hard capacity limit Erlang B formulas can be used to calculate the maximum number of users from offered traffic.

The maximum number of simultaneous users that can have service i are then:

(5)

where

ASHO,i is the offered traffic including soft handover for service i.

KGoS,i is the grade of service in % for service i.

4.4 Calculating number of CEs

4.4.1 nCE,Peak

When the number of required channels per service is known, the number of channel elements to handle peak traffic in busy hour can be calculated:

(6)

where



Mi is the maximum number of simultaneous users per site for service i.

i is the channel element factor for radio bearer.

HSDPA – Dimensioning for UL A-DCH

As mentioned in section 2.1.3, there is no need for dimensioning of channel elements for HS-DSCH and DL A-DCH. However, there is a need for CE dimensioning for UL A-DCH. Equation 6 can be applied for 64 kbps or 384 kbps UL A-DCH, since these channels are treated in the same way as DCHs.

EUL - Dimensioning for E-DCH

Equation 6 can also be used for EUL, but with some adaptation. The reason for the adaptation is different handling of serving cell E-DCH and non-serving cell E-DCH users in soft handover, described in section 4.2.3. As a result, the following expression is used to calculate nCE in equation 6.

(7)

where

MEUL is the maximum number of simultaneous E-DCH users per site.

eulServing is the channel element factor for the target rate serving cell E-DCH users. The channel element factor depends on the setting of the parameter eulTargetRate.

i is the fraction of soft handover users for service i.

eulNonServing is the channel element factor for fixed CE allocation for non-serving E-DCH users in soft handover. The fixed CE allocation depends on the setting of the parameter eulReservedHwBandwidthSchedDataNonServCell.

4.4.2 nCE,AVE

When conversational and best effort bearers are mixed, the dimensioning has to take the average traffic into consideration. With the conversational traffic having higher priority, the interactive traffic will fill up the space when the conversational class traffic is not being used. This can be described by the following figure:

Time

100% Channels available for packet switched traffic

Average circuit switched traffic

Channels available for circuit switched traffic O

ccup

ied

chan

nels

Figure 4. Mixing of circuit and packet switched traffic



This gives two possible limitations, a limitation from conversational peak traffic, and a limitation from best effort traffic, since high data rate services are more demanding in terms of channel element capacity. When estimating the channel elements for best effort traffic both average conversational traffic and best effort traffic have to be taken into account.

The average CE usage for conversational class traffic is calculated with the following equations:

(8)

where

ASHO,i is the offered conversational class traffic in Erlang for service i.

pB,i is the actual blocking for service i.

In case of complete partitioning, pB,i will be equal to the GoS desired for each individual service i. If statistical multiplexing is used, the actual blocking value is calculated according to 4.2.4.

4.4.3 nCE,BE

The average CE usage for best effort traffic is calculated with the following equations:

(9)

where ASHO,i is the offered best effort traffic in Erlang for service i.

Enhanced Soft Congestion

In loaded traffic situations, if there is a request to establish radio bearer with a lower rate, Enhanced Soft Congestion will attempt to downswitch best effort users to a lower rates. That allows the dimensioning calculations to be simplified, assuming that only the lowest bearer rate will be used for best effort services. The following formula can then be applied to express nCE instead of equation 9:

(10)

HSDPA - Dimensioning for UL A-DCH

As mentioned in section 2.1.3, there is no need for dimensioning of channel elements for HS-DSCH and DL A-DCH. However, there is a need of CE dimensioning for UL A-DCH. Equation 9 can be applied for 64 kbps or 384 kbps UL A-DCH, since these channels are treated in the same way as DCHs.

EUL - Dimensioning for E-DCH

Equation 9 can also be used for EUL with some adaptation. The reason for the adaptation is the different handling of serving cell E-DCH users and non-serving cell E-DCH users in soft handover as described in section 4.2. As a result, the following formula can be used to express nCE for EUL in equation 9.



(11)

where

AEUL is the offered uplink traffic in Erlang for EUL/HS.

eulServing is the channel element factor for the target rate for EUL users having the cell as a serving EUL cell. The channel element factor depends on the setting of the parameter eulTargetRate.

i is the fraction of soft handover users.

eulNonServing is the channel element factor for fixed CE allocation for non-serving E-DCH users. The fixed CE allocation here depends on the setting of the parameter eulReservedHwBandwidthSchedDataNonServCell.

In a loaded situation, if there is a request to establish radio bearer with a low rate, Enhanced Soft Congestion will request the EUL scheduler to schedule serving cell E-DCH users to rates lower than eulTargetRate. However, a serving cell E-DCH user which has been admitted, will not be downswitched to rates lower than minimum CE allocation. That allows the following formula to be used to express nCE.

(12)

Where minCeAllocation is the channel element factor for the rate supported by the minimum HW allocation for an E-DCH user. These channel elements will not be targeted by the Enhanced Soft Congestion function.

4.4.4 Finding the total number of CEs required

In order to estimate the total number of CE the average CEs and the best effort CEs are added. This is then compared to the CEs occupied by the peak traffic. The following is thus true for the total number of CEs needed:

(13)

A number of different assumptions and strategies regarding peak traffic in busy hour can be used:

Only conversational traffic admitted in busy hour – this strategy results in the lowest cost in terms of CE usage, but results in a throughput degradation in busy hour for best effort traffic. The following is an example how to use equation 13 with this strategy:

Conversational and best effort traffic admitted in busy hour + maximum impact of Enhanced Soft Congestion – this strategy is more costly in terms of CE usage but results in lower throughput degradation in busy hour for best effort traffic, since CEs are dimensioned to admit best effort users in busy hour on maximum 64/64 kbps. The following is an example how to use equation 13 with this strategy. It should be noted that nCE,AVE and nCE,BE do not



need to be taken into account, since these numbers are always lower than nCE,Peak:

Conversational and best effort traffic admitted in busy hour + no impact of Enhanced Soft Congestion – this strategy is the most costly in terms of CE usage. It does not result in any throughput degradation in busy hour for best effort traffic, since CEs are dimensioned to handle all services and rates in busy hour. The following is an example how to use equation 13 when dimensioning for conversational traffic and best effort traffic in busy hour on three different radio bearers:

4.5 Special traffic cases

In build-up phases, before a significant amount of traffic is present in the system, it may be desirable to dimension for just enough CEs to ensure a nominal level of uninterrupted service and seamless coverage. This case still requires a minimum amount of CEs. The following rule of thumb should be used:

The minimum number of CEs in UL and DL should be able to handle the most demanding CS RAB and the most demanding PS RAB

simultaneously

As an example, assume that speech 12.2 kbps, conversational 64 and interactive PS384/384 RABs is used, with the CE factors 1/1, 4/2 and 16/8 for UL/DL speech, CS64 and PS64/384 respectively. Then the minimum requirement would be 4+16 = 20 CE in the UL and 2+8 = 10 CE in the DL.

5 Examples

5.1 R99, only conversational traffic admitted in busy hour

5.1.1 Purpose

The purpose of this example is to show calculations for a mix of conversational and R99 best effort traffic. This example assumes that only conversational traffic is admitted in busy hour. This means a throughput degradation for best effort traffic.

For simplicity there is only one type of conversational traffic (speech). This example assumes maximum impact of Enhanced Soft Congestion, which means that only 64 kbps radio bearers will be used for best effort traffic. Re-



transmissions are included in the best effort traffic. The grade of Service (GoS) for conversational and best effort traffic is assumed to be 2%.

For simplicity complete partitioning is used, which gives an overestimation of the required resources.

5.1.2 Traffic input

Table 3. Example 1: traffic input

Service RAB RB

Offered traffic per site, A

Activity factor KAF

SHO factor κUL DL

Speech Speech 6 Erlang 6 Erlang n/a 0.3

R99 best effort PS data

PS interactive64, 128,

38450MB 500 MB 0.7 0.3

5.1.3 Identifying the number of simultaneous users

Step 1: Converting best effort traffic to Erlang (equation 2)

A64/64,UL = ((5010000008)/3600)/(640000.7) = 2.5 Erlang

A64/64,DL = ((50010000008)/3600)/(640000.7) = 24.8 Erlang

Step 2: Max traffic (equation 3)

A64/64 = max(24.8, 2.5) = 24.8 Erlang

Step 3: Soft handover factor (equation 4)

ASHO,Speech = 61,3 = 7.8 Erlang

ASHO,64/64 = 24.81,3 = 32.2 Erlang

Step 4: Estimating the maximum number of simultaneous users from offered traffic (equation 5)

MSpeech = ErlangB(7.8, 0.02) = 14

5.1.4 Calculating number of CEs

Step 1a: Calculating nCE,Peak per service (equation 6)

nCE,Speech,UL = 141 = 14 CE

nCE,Speech,DL = 141 = 14 CE

Step 1b: Calculating total nCE,Peak for busy hour (equation 6)

nCE,Peak,UL = 14 + 0 = 14 CE

nCE,Peak,DL = 14 + 0 = 14 CE



Step 2: Calculating nCE,AVE (equation 8)

nCE,AVE,UL = 7.81(10.02) = 7.6 CE

nCE,AVE,DL = 7.81(10.02) = 7.6 CE

Step 3: Calculating nCE,BE (equation 9 and 10)

nCE,BE,UL = 32.24 = 128.8 CE

nCE,BE,DL = 32.22 = 64.4 CE

Step 4: Finding the total number of CEs required (equation 13)

nCE,TOT,UL = max(14, (7.6+128.8)) = 136.4 CE

nCE,TOT,DL = max(14, (7.6+64.4)) = 72 CE

5.2 R99, conversational and best effort traffic admitted in busy hour

5.2.1 Purpose

The purpose of this example is to show calculations for a mix of conversational and R99 best effort traffic on the site. This example assumes that conversational as well as best effort traffic is admitted in busy hour. This gives no degradation of the peak throughput for best effort traffic in busy hour, since no impact of Enhanced Soft Congestion is assumed.

For simplicity there is only one type of conversational traffic (speech). Re-transmission is included in the best effort traffic. Grade of Service (GoS) for conversational and best effort traffic is assumed to be 2%.

For simplicity complete partitioning is used, which gives an overestimation of the required resources.

5.2.2 Traffic input


Service RAB RBTraffic distri-bution


Activity factor KAF

SHO factor κUL DL



PS interactive

64/64 70% 50MB 500 MB 0.7 0.3


PS interactive

64/128 20% 50MB 500 MB 0.7 0.3


PS interactive

64/384 10% 50MB 500 MB 0.7 0.3



5.2.3 Identifying number of simultaneous users

Step 1: Converting best effort traffic to Erlang (equation 2) + best effort traffic distribution

A64/64,UL = ((0.705010000008)/3600)/(640000.7) = 1.7 Erlang

A64/128,UL = ((0.205010000008)/3600)/(640000.7) = 0.5 Erlang

A64/384,UL = ((0.105010000008)/3600)/(640000.7) = 0.25 Erlang

A64/64,DL = ((0.7050010000008)/3600)/(640000.7) = 17.4 Erlang

A64/128,DL = ((0.2050010000008)/3600)/(1280000.7) = 2.5 Erlang

A64/384,DL = ((0.1050010000008)/3600)/(3840000.7) = 0.4 Erlang

Step 2: Max traffic (equation 3)

A64/64 = max(17.4, 1.7) = 17.4 Erlang

A64/128 = max(2.5, 0.5) = 2.5 Erlang

A64/384 = max(0.4, 0.25) = 0.4 Erlang

Step 3: Soft handover factor (equation 4)

ASHO,Speech = 61,3 = 7.8 Erlang

ASHO,64/64 = 17.41,3 = 22.6 Erlang

ASHO,64/128 = 2.51,3 = 3.2 Erlang

ASHO,64/384 = 0.41,3 = 0.5 Erlang

Step 4: Estimating the maximum number of simultaneous users from offered traffic (equation 5)


M64/64 = ErlangB(22.6, 0.02) = 31

M64/128 = ErlangB(3.2, 0.02) = 8

M64/384 = ErlangB(0.5, 0.02) = 3





nCE,64/64,UL = 314 = 124 CE

nCE,64/64,DL = 312 = 62 CE

nCE,64/128,UL = 84 = 32 CE

nCE,64/128,DL = 84 = 32 CE

nCE,64/384,UL = 34 = 12 CE

nCE,64/384,DL = 38 = 24 CE




nCE,Peak,UL = 14+124+32+12 = 182 CE

nCE,Peak,DL = 14+62+32+24 = 132 CE


nCE,AVE,UL = 7.81(10.02) = 7.6 CE

nCE,AVE,DL = 7.81(10.02) = 7.6 CE

Step 3: Calculating nCE,BE (equation 9)

nCE,64/64,UL = 22.64 = 90.4 CE

nCE,64/64,DL = 22.62 = 45.2 CE

nCE,64/128,UL = 3.24 = 12.8 CE

nCE,64/128,DL = 3.24 = 12.8 CE

nCE,64/384,UL = 0.54 = 2.0 CE

nCE,64/384,DL = 0.58 = 4.0 CE


nCE,TOT,UL = max(182, (7,6+90.4+12.8+2.0)) = 182 CE

nCE,TOT,DL = max(132, (7,6+45.2+12.8+4.0)) = 132 CE

5.3 R99 + HSPA, only conversational traffic admitted in busy hour

5.3.1 Purpose

The purpose of this example is to show CE calculations for HSPA. For simplicity it is assumed that there is no R99 best effort traffic on the site. UL E-DCH traffic has not been specified. This example assumes that only conversational traffic is admitted in busy hour. This gives a throughput degradation for best effort traffic in busy hour. The example assumes no impact of Enhanced Soft Congestion.



5.3.2 Traffic input


Service RAB RBTraffic distri-bution


Activity factor KAF

SHO factor κUL DL



PS interactive

64/HS 60% 50MB 500 MB 0.7 0.3


PS interactive

384/HS 15% 50MB 500 MB 0.7 0.3


PS interactive

EUL/HS 25% 50MB 500 MB 0.7 0.3

HS target rate 500 kbps

EUL target rate 128 kbps (8CE)

Reservation of CE for EUL SHO 128 kbps (8CE)

5.3.3 Identifying the number of simultaneous users

Step 1: Converting best effort traffic to Erlang (equation 2)

ADL,HS = ((50010000008)/3600)/(5000000.7) = 3.2 Erlang

Step 2: UL A-DCH & EUL traffic distribution

A64HS,UL = 0.750.83.2 = 1.9 Erlang

A384HS,UL = 0.750.23.2 = 0.5 Erlang

AEUL = 0.253.2 = 0.8 Erlang

Step 3: Soft handover factor (equation 4 and section 4.2.3 )

ASHO,Speech = 61.3 = 7.8 Erlang

A64HS,UL = 1.31.9 = 2.5 Erlang

A384HS,UL = 1.30.5 = 0.65 Erlang

AEUL,Serving = 10.8 = 0.8 Erlang

AEUL,nonServing = 0,30.8 = 0.24 Erlang

Step 4: Estimating the maximum number of simultaneous users from offered traffic (equation 5 and section 4.2.3 )


M64HS,UL = ErlangB(2.5, 0.02) = 7

M384HS,UL = ErlangB(0.65, 0.02) = 4

MEUL,Serving = ErlangB(0.8, 0.02) = 4

MEUL,nonServing = ErlangB(0.24, 0.02) = 3








nCE,Peak,UL = 14+0 = 14 CE

nCE,Peak,DL = 14+0 = 14 CE


nCE,AVE,UL = 7.81(10.02) = 7.6 CE

nCE,AVE,DL = 7.81(10.02) = 7.6 CE

Step 3: Calculating nCE,BE (equations 9 and 11)

nCE,64HS,UL = 2.54 = 10

nCE,384,UL = 0,6516 = 10.4

nCE,EUL,Serving = 0,88 = 6.4

nCE,EUL,nonServing = 0,258 = 2


nCE,TOT,UL = max(14, (7,6+10+10.4+6.4+2)) = 36.4 CE

nCE,TOT,DL = max(14, (7,6+0)) = 14 CE

6 CE Observables

6.1 Introduction

The aim of this chapter is to describe a method how to evaluate CE utilization per site based on observables. The chapter also presents a method how to estimate softer handover factors from counters or other observables that can be used when dimensioning channel elements.

6.2 CE utilization using observables

Because the CEs are node specific, it is important to check the Performance Management (PM) statistics for each node before activating license control. Great care must be taken to set the right levels for each individual node.

In order to assess the needs of the system in terms of licenses per node, an evaluation method has been devised., The method to evaluate CE utilization is shown in the following steps:



Activate the RBS counters pmNoOfRadioLinksSf4 to pmNoOfRadioLinksSf256 in both uplink and downlink [4].

Apply formulas for uplink and downlink Channel Element utilization, based on RBS counters.

The formulas for uplink CE (Channel Element) and downlink CE utilization are defined as follows:

fUL(t) = KSHO,UL*(SF256*pmNoOfRadioLinksSf256[t] ++ SF128*pmNoOfRadioLinksSf128[t] + + SF64*pmNoOfRadioLinksSf64[t] + + SF32*pmNoOfRadioLinksSf32[t] ++ SF16*pmNoOfRadioLinksSf16[t] + + SF8*pmNoOfRadioLinksSf8[t] ++ SF4*pmNoOfRadioLinksSf4[t] ) (14)

fDL(t) = KSHO,DL *(SF256*pmNoOfRadioLinksSf256[t] + + SF128*pmNoOfRadioLinksSf128[t] ++ SF64*pmNoOfRadioLinksSf64[t] ++ SF32*pmNoOfRadioLinksSf32[t] ++ SF16*pmNoOfRadioLinksSf16[t] ++ SF8*pmNoOfRadioLinksSf8[t] ++ SF4*pmNoOfRadioLinksSf4[t]) (15)

where KSHO,UL and KSHO,DL is the softer handover factor, which compensates for the fact that the cost-model is defined for radio link sets, while the counters observe the number of radio links.

The formulas can be plotted as graphs, which can give an indication if the total CE license should be increased.

The counters to observe belong to the MO Classes UplinkBasebandPool, and DownlinkBasebandPool respectively. Noted that these counters measure the number of radio links (sampled every minute in a histogram (ddm) counter) on the whole RBS. This means that the values observed here include soft handover and softer handover.

Softer handover does not need to be considered, since it does not influence the amount of consumed CEs. Therefore it is required to compensate for softer handover when using Radio Link based counters to observe consumed CEs. The compensation is done with the factors KSHO,UL and KSHO,UL. These factors range typically around 0.9 but can also be estimated. Section 6.3 shows a method to estimate them.

6.2.1 CE observability for Enhanced Uplink

The RBS counters pmNoOfRadioLinksSf4 to pmNoOfRadioLinksSf256 take EUL links with actual rate into account. The method for monitoring the total CE utilization can thus also be applied on RBSs supporting EUL. The RBS counter pmHwCePoolEul is used to monitor the total sum of CEs allocated on the UL hardware by the EUL scheduler.



6.3 Estimation of KSHO,UL and KSHO,UL

The aim of this section is to present a method how to estimate KSHO,UL and KSHO,UL. These factors can be expressed as the ratio between the number of radio link sets and the number of radio links on a site. Table 6 shows the counters to use to calculate the number of radio link sets and the number of radio links [5].

Table 6. Counters for handover performance

Counter Radio links and sets

A pmSumUesWith1Rls1RlInActSet Single link

B pmSumUesWith1Rls2RlInActSet Softer handover (2 links)

C pmSumUesWith1Rls3RlInActSet 3 way softer handover (3 links)

D pmSumUesWith2Rls2RlInActSet Soft handover (2 links)

E pmSumUesWith2Rls3RlInActSet Soft + softer (3 links)

F pmSumUesWith2Rls4RlInActSet Soft + 2 (or 3) way softer (4 links)

G pmSumUesWith3Rls3RlInActSet 3 way soft (3 links)

H pmSumUesWith3Rls4RlInActSet 3 way soft + softer (4 links)

I pmSumUesWith4Rls4RlInActSet 4 way soft (4 links)

The counters above show how many radio links and radio link sets that have been active per cell during the counting reporting period. To obtain KSHO,UL and KSHO,UL values for the entire site the values for the individual cells belonging to the site are summed up. For UEs in softer handover (B, C, E, F and H) there is a need to weigh before summarizing the counter values. The weighting is done to compensate for the fact that softer handover UEs will be counted in several cells on a site. Figure 5 and Figure 6 give examples of weighting for the most common softer handover cases.

Figure 5. Example of weighting factor for 2 way softer handover and 3 way softer handover

In example 1 in Figure 5, pmSumUesWith1Rls2RlInActSet will be incremented in both cell A and B. Therefore the value of pmSumUesWith1Rls2RlInActSet needs to be weighted to reflect number of radio link sets on that site. The weighting factor for pmSumUesWith1Rls2RlInActSet is ½, since the counter is stepped in 2 cells.

In example 2 pmSumUesWith1Rls3RlInActSet will be incremented in cell A, B and C. The weighting factor for pmSumUesWith1Rls3RlInActSet is 1/3, since the counter is stepped in 3 cells.


A

B

A

B

A

B

C

A

B

C

Example 1 Example 2


Example 3 in Figure 6 shows 2 different cases where pmSumUesWith2Rls3RlInActSet will be incremented. As a result of that the following assumption is made:

50% probability that case 1 is valid

50% probability that case 2 is valid.

In case 1 the counted value of pmSumUesWith2Rls3RlInActSet is weighted with a factor ½, since the counter is incremented in 2 cells belonging to site A. In case 2 the counted value of pmSumUesWith2Rls3RlInActSet is weighted with a factor 1, since the counter is only incremented in one cell belonging to site A. With the assumption made the weighting factor for pmSumUesWith2Rls3RlInActSet is 1/2(1/2+1) = 3/4.

Figure 6. Example of Soft + softer (3 links) handover

Table 7 shows weighted counter values for all cases involving softer handover. Note that the weighting factors have been estimated under certain assumptions, since it is not possible to calculate the weighting factors exactly to reflect all situations in a radio network

Table 7. Weighted counters for softer handover

Radio links and sets Counter and weight

b total number of softer handover (2 links) UEs.

pmSumUesWith1Rls2RlInActSet 1/2

c total number 3 way softer handover (3 links) UEs.


e total number of soft + softer (3 links) UEs.


f total number of soft + 2 (or 3) way softer (4 links) UEs.


h total number of 3 way soft + softer (4 links) UEs.


When the number of active radio links and radio link sets have been are estimated, the softer handover factors are calculated with the following equation:


A

B

C

Site A

Case 1

A

B

C

Site A

Case 2

Example 3

Site B Site B


(16)

where the quantities Ai to Ii and bi to hi are defined by Table 6 and Table 7

7 References1. RBS 3000 Baseband Product Description, 1/221 01-FGC 101 811

2. User Description EUL Scheduler, 112/1553-HSD 101 02/5

3. Statistical Multiplexing, Theory and Implementation, ERA/FN/R-01:0087

4. User Description Performance Statistics, 79/1551-HRB 105 102/1 Uen

5. User Description Performance Statistics, 58/1551-AXD 105 03/1 Uen


channel element dimensioning guideline

Documents

channel element resource

ericsson ab

conversational traffic

number of users

channel element cost

busy hour

basic resource model

best effort traffic