Power Management for Throughput Enhancement in Wireless Ad-Hoc Networks using Clusters

Power Management for Throughput Enhancement in Wireless Ad-Hoc Networks using Clusters

Abstract

Mobile Adhoc network is an autonomous network of mobile nodes where the mobiles uses the services of other mobiles for routing and packet transmission.

In this project we introduce the notion of power management within the context of wireless ad-hoc networks. More specifically , we investigate the effects of using different transmit powers on the average power consumption and end-to-end network throughput in a wireless ad-hoc environment. This power management approach would help in reducing the system power consumption and hence pro-longing the battery life of mobile nodes. Furthermore, it improves the end-to-end network throughput as compared to other ad-hoc networks in which all mobile nodes use the same transmit power. Further the work is extended to cluster based routing where the data is routed through higher energy nodes called clusters with dynamic cluster management.The improvement is due to the achievement of a tradeoff between minimizing interference ranges, reduction in the average number of hops to reach a destination, the probability of having isolated clusters, and the average number of transmissions (including retransmis-sions due to collisions). The protocols would first dynamically determine an optimal connectivity range wherein they adapt their transmit powers so as to only reach a subset of the nodes in the network. The connectivity range would then be dynamically changed in a distributed manner so as to achieve the near optimal throughput. Minimal power routing is used to further enhance performance. Simulation studies are carried out in order to investigate these design approaches using Omnet++ event driven simulator. It is seen a network with such a power managed scheme would achieve a better throughput performance and lower transmit power than a network without such a scheme.

Chapter 1: Preamble1.1 Statement of the Problem The fundamental problem with Adhoc network are mainly twofold:1) No control over the sessions due to decentralized topology2) No power control technique due to non centralized architecture.Various techniques are being proposed to solve the problem, mainly by using Backbone routing, cluster based routing and so on. Various power control algorithms are also implemented for enhancing the energy utilization of the network. In this work we combine the power control and cluster based routing features to derive a new routing and transmission protocol for manet to improve the performance of the system.1.2 Objective of the StudyVarious routing algorithms have been proposed for wire-less ad-hoc networks in the literature. Those algorithms are mainly focused on establishing routes, and maintain-ing these routes under frequent and unpredictable connectivity changes. The implicit assumption in most of the earlier work is that nodes' transmitted powers are fix ed. To the best of our knowledge, there is no prior known work that proposes the concept of mobile ad-hoc nodes using different transmit powers. It is evident that this approach is restricted to ad-hoc networks of relatively low mobility patterns. If the nodes are highly mobile, the power management algorithm might fail to cope with the fast and sudden changes due to fading and interference conditions.therefore the objective is to investigates the benefits, and possibly the tradeoffs, of deploying different transmit powers in the wireless ad-hoc environment. We propose a power management scheme with cluster based topology control which can be used in conjunction with traditional table-driven routing protocols, with possibly minor modifications. The performance measures are taken to be the end-to-end network throughput and the average power consumption.1.3 Scope of the Study The work can be used as a benchmark for practical Adhoc networks like office network which forms clusters of mobile nodes with laptops as cluster heads or domain controllers. The system can be used to enhance the lifetime of the nodes by suitable power management. The system produces better performance in terms of higher packet delivery ratio and throughput. The load is distributed through cluster heads which are essentially more energy efficient nodes. The system can also be used for transmission of multimedia traffic or VBR traffic where more stability is needed for the topology for data transmission.1.4 Introduction to Mobile Adhoc Network[Yourself]

1.5 Limitation of the Study Even though the system produces better performance than conventional AODV based techniques, the protocol applies higher load on cluster nodes. If the cluster nodes are considered to be normal mobiles and not Laptops or other power enabled nodes, then the system drains the node's energy thick and fast. Further the power management process introduces higher latency to the system.

Chapter 2

Related Work

[Yourself]

Chapter 3Power Management in MANETWhen the power management scheme is implemented, each node transmits at a minimum power level such that only a fix ed number of neighboring nodes can hear the transmission. For example, a node might transmit with a power such that only its three closest neighbors can hear its transmission.Thus, in Figure 1 below, node A transmits with a power P1 such that only it's three nearest neighbors i.e., nodes B, C and D can hear it. Similarly, node D would transmit with a different power, say P2 such that only it's three nearest neighbors i.e., nodes A, C and E can hear it.

In order to set up the framework to investigate the effect of transmit powers on the end-to-end network throughput, we make the following assumptions and introduce some appropriate notations:

1. The wireless ad-hoc network consists of n nodes; each node has a unique ID, denoted by Node ID.

2. The mobile nodes are assumed to have low mobility patterns, that is, they are typical pedestrians. This, in turn, implies that the network topology changes slowly and the class of shortest-path routing algorithms is applicable. 3. Each mobile node has direct connectivity to its N clos-est neighbors only, where N is to be adapted dynamically.

4. Assume connectionless (datagrams) type of traffic, i.e. routing decisions are made on a packet-by-packet-basis.

The transmit power of any mobile node is upper bounded by a maximum power level denoted as Pmax The limited size and weight of the mobile terminal dictate this constraint.

The transmit power of any mobile node is lower bounded by a minimum power level Pmin. This constraint is essential to avoid partitioning the network into isolated islands. 1.1GeneralThe section describes the procedures for the support of the service-level interworking for the Short Message Service as defined in [C.S0015] and Instant Messaging service as defined in [OMASIMPLE]. NOTE:In the procedures in the following subclauses, the ICSCF, PCSCF and ASs such as IM AS are not shown in the figures.1.2SMS-GW IMS 3rd Party Registration call flowThis call flow shall be according to the procedures described in [X.S0048].1.3Interaction between transport-level and service-level interworking with interworking in the originating side1.3.1GeneralThe interaction between transport-level interworking (between SMS over CS and SMS over IMS) and service-level interworking (between Instant Messaging and SMS) depends on the user subscription and authorisation, on the UE capabilities, and on operator policy.If a user is only subscribed to either transport-level interworking or service-level interworking, only procedures defined for the subscribed interworking type may be performed.If a user is subscribed to both transport-level interworking and service-level interworking, but the user is only authorized for one of the interworking types when the message is processed, only the authorized interworking may be performed.If a user is subscribed to both transport-level interworking and service-level interworking, and the user is authorized for both types, the behavior of the SMS-GW depends on the specific scenario, on the registered capabilities of the UE, and finally is defined by operator policy and user preferences.For a user subscribed to service-level interworking, two Application Servers in the network are normally called upon to handle an Instant Message:-the IM AS, defined in [OMASIMPLE]; -the SMS-GW.The following sections describe the different interaction scenarios.1.3.2IMS OriginatingIn the originating network, a UE sends a SIP MESSAGE (Encapsulated Short Message or Instant Message). The originating SCSCF forwards the SIP MESSAGE to the SMS-GW based on the iFC. If there is no subscription for the interworking service, the SMS-GW is not included in the iFC and the SCSCF continues with the subsequent iFC check. After all the originating iFC triggers have been handled, the SCSCF attempts to route the SIP MESSAGE to the terminating IMS network. If it fails, an error is returned to the sender.NOTE1:If an IM AS is present in the network, Instant Messages are routed to it before going to the SMS-GW.NOTE2:An encapsulated Short Message uses the PSI of the Message Center as the Request-URI. If the user is not subscribed to transport-level interworking and the SMS-GW is not invoked, the ENUM query fails, and an error is returned to the sender. How the UE is provided with the PSI of the Message Center is outside the scope of this document.When the SMS-GW receives the SIP MESSAGE, it shall decide which interworking is performed based on the content of the received SIP MESSAGE, as the SMS-GW can distinguish between an encapsulated Short Message and an Instant Message. If an encapsulated Short Message is received and if the subscriber is authorized for the transport-level interworking, the SMS-GW maps the encapsulated Short Message to a Short Message. Similarly, when an Instant Message is received, the SMS-GW determines whether the Instant Message is routable in IMS. If the Instant Message is not routable in IMS and the service level interworking is authorized, the SMS-GW shall perform the service-level interworking. Figure 1 Figure 6.3.2: Performing interworking service on originating side1.3.3IMS TerminatingWhen the SMS-GW receives a Short Message from the legacy network on the terminating side, it performs the domain selection to determine the preferred domain to transfer the Short Message. If the selected network is IMS, the SMS-GW will determine whether the transport level interworking or the service level interworking is to be preformed based on the users' subscription and authorization, and on the UE capability as indicated during IMS registration. If the user has subscribed to both services, is authorized for both and the UE has indicated its capability to receive both encapsulated Short Messages and Instant Messages, the priority between the transport-level interworking and the service-level interworking is based on operator policy and user preferences.

NOTE:If the incoming Short Message is interworked to an Instant Message, the resulting Instant Message could be routed to the IM AS before being sent to the UE. Figure 2 Figure 6.3.3: Performing interworking service on terminating side for an incoming Short Message1.4IM capable UE sends an Instant Message to an SMS user with interworking in the originating side Figure 3 Figure 6.4: Successful IM origination to SMS procedure1)The UE registers to S-CSCF according the IMS registration procedure.2)UE submits the Instant Message to the S-CSCF using an appropriate SIP method. The UE may request to hide its Public User Identity from the recipient within the Instant Message, as described in [OMASIMPLE].3)S-CSCF forwards the Instant Message toSMS-GW based on stored iFC.NOTE 1:Subscribers with no subscription for service level interworking will not be provided with the relevant iFCs.4)The SMS-GW shall decide whether to perform service-level interworking depending on SIP request header field (e.g. Request-URI), operator policy, when the Instant Message is not routable in the IMS. If the service-level interworking is authorized, the originating UE's SMS-GW delivers the SMS message to the terminating SMS-GW in a MAP SMDPP message. The terminating SMS-GW is not shown for brevity. 5)The terminating SMS-GW responds by sending a MAP smdpp message back to the sender of the MAP SMDPP message. 6)If service authorization is successful, the SMS-GW acknowledges the Instant Message.7)Instant Message acknowledgement is forwarded by S-CSCF to UE.NOTE2:Steps 6 and 7 can occur anytime after the subscriber authorization check has been performed by the SMSGW.8)The terminating SMS-GW acknowledges message delivery to the MS by sending MAP: SMDPP (Delivery Report).9)The originating SMS-GW responds by sending a MAP smdpp message back to the sender of the MAP SMDPP message. 10)SMS-GW translates the received Delivery report to an appropriate Instant Message, and forwards it to the SCSCF. If the SMSGW sent concatenated Short Messages to terminating SMS-GW in step4, the SMSGW should wait for the last Delivery Report, and translate the last Delivery Report to an appropriate Instant Message, and forward it to the SCSCF.11)S-CSCF sends the translated Instant Message to the UE.12)UE acknowledges the translated Instant Message.13)Acknowledgement of the translated Instant Message is forwarded by S-CSCF to SMS-GW.1.5IM capable UE sends an Instant Message to an SMS user with interworking in the terminating sideThis procedure describes the delivery of an Instant Message to a registered IMS subscriber that is presently being served by a 1xRTT network. Figure 4 Figure 6.5: Successful IM terminating to SMS procedure with interworking in the Terminating Side1)UE submits an Instant Message, destined to another IM user in another IMS domain, using an appropriate SIP method. The UE may request to hide its Public User Identity from the recipient within the Instant Message, as described in [OMASIMPLE].2)The S-CSCF resolves the destination domain and routes the message towards the S-CSCF in the terminating network ("Terminating S-CSCF").3)The terminating S-CSCF forwards the Instant Message to the IM AS ("Terminating IM AS") based on stored iFC.NOTE:Depending on iFC configuration, it is possible that the IM AS is not triggered for the unregistered subscribers.4)The terminating IM AS invokes terminating IM services as applicable for the destination IM user. 5)The IM AS can forward the Instant Message back to the terminating S-CSCF, e.g. when the terminating IM user is offline.6)The terminating S-CSCF forwards the Instant Message to the SMS-GW, e.g. based on stored iFC.7-11)The SMS-GW sends Accepted towards the IM capable UE to indicate that the Instant Message has been accepted for further processing.12)The SMS-GW performs service level interworking of the received instant message (step 11). After the service level interworking, the SMS-GW sends a MAP SMDPP Invoke message to the Serving MSC and starts timer SMT. The MAP SMDPP Invoke message containing the SMS Delivery message [C.S0015] in the SMS_BearData Parameter.13)The MSC sends an ADDS Page message [A.S0014] to the BS. The ADDS Page message contains the SMS Delivery message in the ADDS User Part information element.If the MSC requires an acknowledgment, it includes the Tag information element in the ADDS Page message and starts timer T3113.14)The BS sends the SMS Delivery Message to the MS on the Paging Channel or the Forward Common Control Channel. Before sending the short message, the BS may perform vendor specific procedures such as paging the MS to determine the cell in which the MS is located.15)If a Layer 2 Ack was solicited in the Data Burst Message (Step 14), the MS acknowledges the receipt of the message by a Layer 2 Ack.16)If the MSC requested an acknowledgment by including the Tag information element in the ADDS Page message (step 13), the BS replies with an ADDS Page Ack message including the Tag information element set identical to the value sent by the MSC (step 13). If timer T3113 was previously started, it is now stopped.17)The MSC acknowledges the MAP SMDPP invoke message (step 12) by sends a SMDPP return result to the SMS-GW. Upon receiving the MAP SMDPP return result message the SMS-GW stops timer SMT.18)If a Reply Option subparameter received in an SMS Deliver Message (step 14) indicates that User Acknowledgment is requested, the mobile station should indicate the request to the user. When the user acknowledges the message, the mobile station sends an SMS User Acknowledgement Message in response to the SMS Deliver Message.19)If a Layer 2 Ack was solicited in the Data Burst Message (Step 18), the BS acknowledges the receipt of the message by a Layer 2 Ack.20)The BS sends the MSC an ADDS Transfer message. The ADDS Transfer message contains the SMS User Acknowledgment Message in its ADDS User Part information element.21)The MSC sends the SMS-GW a MAP SMDPP Invoke message and starts timer SMT. The MAP SMDPP Invoke message contains the SMS User Acknowledgment Message in the SMS_BearData Parameter.22)The SMS-GW acknowledges the MAP SMDPP invoke message (step 21) by sending an SMDPP return result to the MSC. Upon receiving the MAP SMDPP return result message the MSC stops timer SMT.23)SMS-GW translates the received SMS User Acknowledgment Message to an appropriate Instant Message, and forwards it to the terminating SCSCF.24-27) The terminating S-CSCF sends that Instant Message containing the delivery status of the message towards the IM capable UE.28-32) The IM capable UE sends OK response the SMS-GW.1.6IM user receives Short Message from an SMS userAn IMS registered user with SIMPLE IM service receives a Short Message formatted via service-level interworking to an Instant Message. We deploy the classical shortest-path routing algorithm with a slight modification where routing is performed through cluster heads. The cluster heads are selected absed on the degree of neighborhood and energy consideration. The link costs are chosen to be the transmitted powers. Therefore, the objective is to route the packet from the source to the destination through the minimum power path.

12.The received power at any mobile node has to be greater than a minimum power level, denoted by Min-RecvPower. This is crucial in order to guarantee reliable communication between the transmitter and the receiver. This value helps determine the power level at which a mo-bile has to transmit in order to directly reach a neighboring node.

13.It is expected that unidirectional links will be formed when transmit powers are thus manipulated, i.e. we might have a cluster of nodes that can communicate with each other but no packets can either enter or leave this cluster. Modifications to the the protocol to eliminate this effect

are being investigated. However, it should be noted that this does not change the routing methodology1 since the signaling channel is bidirectional. 14.The Signaling Packet format is shown in Figure 2 below,

Node IDNeighbor IDTransmit Power Level Figure 2: Signaling Packet Format

where,

Node ID: Identifier for the node broadcasting the signaling packet.

Neighbor ID: Identifier for a direct neighbor to which the node is broadcasting the signaling packet.

Transmit Power Level: Minimum power level needed to reach that neighbor.

15. The Data Packet format is shown in Figure 3.

Source IDDestination IDCurrent Node ID Next Node IDRe-TransmissionsPayload

Figure 3: Data Packet Format

where,

Source ID: Identifier of the node that generated the packet. Destination ID: Identifier of the packet's destination node. Current Node ID: Identifier of the relay node at which the packet is currently stored on its path to the destination.

Next Node ID: Identifier of the next relay node to which the packet is to be transmitted on its path to the destina-tion.

Re-Transmissions: Total number of retransmission at-tempts performed on that packet. (retransmission will be17. Node Throughput is defined as percentage of success-ful transmission attempts.

End-to-End Network Throughput is defined as per-centage of packets that reach their destinations success-fully

The power model is given by

where Pr is the received power, and Pt is the transmission power and is the interference ratio.The Minimum Power Routing (MPR) algorithm proposedis a hop-by-hop shortest path routing mechanism where the link costs are the transmitted power levels.The routing algorithm then goes through the followingsteps:1. Based on the routing table constructed, the mobile nodecreates the set of all possible routes from the source to destination.2. The routing algorithm employed falls within the general class of shortest path routing. It searches, within the created route set, for the minimum cost route from sourceto destination.3. Determine the next relay node on the minimum power route.

4. Modify the Next Node ID field in the data packet being routed.

5. Copy the packet to the retransmission buffer until its successful reception at the next node is indicated via an ACK message.

6. The packet is sent to the MAC module for transmission to the next relay node.

Connectivity ProblemConsidering various conditions of connectivity and power management, it is straightforward to point out the follow-ing issues:Consider a wireless ad-hoc network with all mobile nodes using the maximum power level (i.e. no power man-agement). Accordingly, any mobile node can reach a large number of nodes in just one hop. The advantage of this ap-proach is reaching a large number of nodes in a single hop and almost all of the nodes in the network in two hops. The price paid is however twofold, namely high power con-sumption and higher interference, which results in a large number of collisions. If the link cost is taken to be the transmitted power, it is straightforward to notice that the Cost of the Links are equal to Pmax

Each mobile node has a direct link to the closest N out of (n-1) mobile nodes. We call these N nodes a cluster. Given N, the mobile node adjusts its power to reach at most the farthest node within its cluster. However, we assume that there is no power adaptation within the cluster. The advantages of this approach are lower power consumption and possibly, a node's transmission will cause lower interference to other simultaneous transmissions, when com-pared to the previous case. The drawbacks are a higher number of hops might have to be traversed in order to reach a destination, and there exists the possibility of having iso-lated clusters. Note that link costs (transmitted powers), in this context, are generally different depending on the ra-dius of each cluster. Accordingly, incorporating the minimum power routing algorithm is crucial to limit power consumption.

Power Measurement

This procedure emulates the operation of mobile node j capturing the beacon signal transmitted by node i during node i's allocated signaling slot, where and i!=j. The received signal strength depends solely on the transmitted power level (which is assumed to be Pmax during this phase), the current positions of nodes i and j, and the effect of the log-normal shadowing. Thus, the received power level is computed by using the following formula:

As pointed out earlier, we rely on average power measurements rather than instantaneous power measurements. This is due to the fact that instantaneous measurements could be inaccurate in reflecting the slowly varying channel conditions in the presence of fast multipath fading. Therefore, a moving average is computed by each node to average out the fast fading over a pre-specified number of most recent instantaneous power measurements.

Power Management

There are two suggested approaches for power manage-ment in mobile ad-hoc networks:

No power adjustment within a cluster. Power adjustment within a cluster.

The basic difference between the two schemes is that in the former scheme, the power needed to communicate with the farthest node in the cluster is also used to communi-cate with any closer node in the cluster. On the other hand, the latter scheme suggests communicating with each node using the minimum power it needs for reliable communi-cation. This introduces less interference to simultaneous transmissions of other nodes.

The objective of defining a cluster is to reduce colli-sions/interference and thereby improve the end-to-end net-work throughput. As mentioned earlier, we assume a minimum required level of received power, denoted Min-RecvPower, that is necessary to guarantee a maximum ac-cteptable bit error rate. The minimum power level to be transmitted by node i such that at least the MinRecvPower level is achieved at node j for a given network configura-tion is given by:

Chapter System DesignData Flow Diagram of the Technique

Forward if Current Node is CHControl Power based on BERPower Measure

Sequence Diagram

Chapter Implementation. Cluster formation using a. degree b. energyn handleHello method whenever a helloPacket Comes Energy will be retrived from physical layer, Cluster Head CalculationDegree of Neighborhood should be high,m Entropy, Low, Mobility low

evnumHost;cModule* sm = (cModule*) simulation.module(parentModule()->id());aodv*s1=(aodv*)sm->submodule("route");int cnt=0;/*for( cQueue::Iterator iter(s1->routeTab,1) ; !iter.end(); iter++){cnt++;}ev