Performance Analysis of Power -aware Node-disjoint Multipath Source Routing in Mobile Ad Hoc Networks

Transcription

1 2017 IEEE 7th International Advance Computing Conference Performance Analysis of Power -aware Node-disjoint Multipath Source Routing in Mobile Ad Hoc Networks M.Bheemalingaiah 1, M. M. Naidu 2 Dept. of CSE Dept. of CSE Ellenki College of Engineering and Technology S.V. University Hyderabad, India Tirupati, India saibheem2008@gmail.com mmnaidu@yahoo.com D. Sreenivasa Rao 3 P.Vishvapathi 4 Dept. of ECE Dept. of CSE J.N.T. University Ellenki College of Engineering and Technology, Hyderabad, India. Hyderabad, India dsraoece@gmail.com vpujala@gmail.com Abstract: Since 2000, Mobile Ad Hoc Networks are treated as the emerging filed in the wireless communication. They comprise only mobile nodes that use wireless transmission and can be set anywhere and anytime because they eliminate complexity of infrastructure and central admission. The Mobile Ad Hoc Networks are extensively used in different fields such as emergency situations, military applications and mobile communications. The routing is the major issue in the field of MANET due to the mobility nature and lack of infrastructure of the network. The different routing protocols have been proposed to address the routing issue. The development of multipath routing for mobile ad hoc network by considering the performance metrics with standard simulator is an important research area. Hence in this paper, we have chosen Power-aware Node-disjoint Multipath Source Routing (PNDMSR) to implement and analyze its performance with respective to Multipath Dynamic Source Routing (MDSR) by using various quantitative performance metrics like, routing control overhead, throughput, packet delivery ratio, packet loss and energy efficiency by varying various parameters like network s size, mobility of node, pause time, data rate and load.the main objective of the PNDMSR is selecting energy aware node-disjoint multipath from source to destination by optimizing the overhead using node s cost and it increases the network of lifetime. Keyword MANET, PNDMSR, MDSR, Energy. Routing, Overhead. I. INTRODUCTION MANETs are considered as autonomous, self-configured, multi-hop wireless networks. They don t rely on any stationary infrastructure, no centralized control and they are quickly deployable anywhere at any time. In the MANET, all nodes are mobile that cooperate in friendly manner, they are connected dynamically and they have the ability to leave existing nodes from the network and enter new nodes into the network at any time, due to this, the topology of the MANET is highly dynamic and it frequently changes. In the MANET, every node the route discovery and route maintenance phase. The MANEs are widely used in military, civilian and commercial applications [1][2].The most of existing routing protocols are single path routing protocols. They find single route and utilize it for data transmission from source to destination. Due to dynamic topology, poor and variable wireless links, dynamic characteristics of radio channel, node failure, the currently using route becomes invalid. Due to this the overhead for finding new route form source to destination may be high and extra delay for new route discovery may be introduced and data transmission becomes late [3]. To overcome this drawback by using multipath routing, the multipath routing is latest trend in the MANET; it finds multiple paths from any source to destination in a single route discovery. Due to the introduction of multipath routing, the time for searching the route will be reduced drastically and also it helps to reduce the latency for searching the other route at the situation of path failure. In MANETs, multipath routing is considered as trusted approach for ad hoc networking. Due the mobile nature or weak signal strength, the node failure occurs frequently in the MANETs. At the condition of route failure, the multipath routing serves as the best approach for finding the alternate path. The applications of multipath routing are given below. Reliability Fault tolerance Balancing Of Load Bandwidth aggregation Reduced delay /17 $ IEEE DOI /IACC

2 Energy Consumption Reduction of Routing Overhead Provide Quality of Service Improve Network Security and Secure communication In the recent years, many researchers made the contribution to address the multipath routing. Some of the well known and standard multipath routing protocols are AODVM [4], AOMDV [5], SMR [6] and MSR [7].The MANETs contains only mobile modes; they are originally operated with the battery power, so, the major constraint for mobile nodes are battery power. Therefore, the energy is consumed during the transmission of packets. Hence energy optimization is another important research challenge [1]. Optimizing the energy consumption is one of the designing issues for multipath routing. To serve this purpose, several energy aware multipath routing protocols were proposed [2]. The optimization of routing overhead is crucial part when designing new multipath routing protocol because the overhead directly is proportional to congestion, probability of collisions, packet loss, packet error rate, packet delay, bandwidth consumption and energy consumption. It is indirectly proportional to packet delivery ratio and throughput. Hence the overhead depends on overall performance of the multipath routing protocol. The Power-Aware Node-disjoint Multipath Source Routing (PNDMSR)[8] protocol is significantly different in the way approach for finding energy aware node-disjoint multipath from source to destination, because it uses low overhead broadcast technique to optimize the overhead in different cases. Hence we select the PNDMSR [8] for analysis of its performance and to prove it is efficient, scalable and robust. II. PNDMSR and MDSR Protocols The MDSR is termed as Multipath Dynamic Source Routing which is an extension of Dynamic Source Routing (DSR) [9] it one of well known standard single path routing protocol. The functionality of MDSR same as functionality of Multipath Source Routing (MSR) [7] and it resembles the nature of MSR [7]. The MDSR is considered as candidate routing protocol to propose the PNDMSR [8] to find energy aware node-disjoint multipath form source to destination by performing several modifications in MDSR. The following five mathematical models have been proposed in the PNDMSR [8] Network Model Energy Consumption model Multipath Node-disjoint Model Model of Path Failure Path Selection Model In the PNDMSR, the cost function is assigned each node. It is also called node s cost and it based residual energy of node. The objective of this cost function is to select the optimal energy aware node-disjoint multipath path from source to destination and increase the life time of network. The four main mechanisms which serve the PNDMSR are route discovery, route selection, route utilization and route maintenance. A. Route Discovery in PNDMSR In the PNDMSR, it employs energy aware optimal nodedisjoint multipath from source node to destination node by optimizing routing overhead. Two novel techniques were proposed to optimize the routing overhead [8].First novel technique is broadcast with low overhead approach; the main objective of this approach is to optimize the flooding of packets during route discovery. In this approach, during flooding of RREQ (Route Request) packets from source to destination, many duplicate RREQ packets are eliminate by each intermediate node itself by using path s cost. Each RREQ packet carries the path and its cumulative cost. Second novel approach is Computation of optimal and feasible energy aware node-disjoint multipath is done by destination node. In the PNDMSR, The destination sends back multiple Rout Reply (RREQ) packets to source. Each RREQ packet carries the path and its cumulative cost. In the traditional energy aware multipath routing, the source node computes energy aware multipath from itself to destination during route discovery. If the destination sends all energy aware multiple paths in the Route Reply (RREP) packets to source via intermediate node then overhead is increased from destination to source. In the PNDMSR eliminates some unnecessary multiple paths that have less energy (high cost) in the route reply. The computation of the energy aware node disjoint optimal paths and feasible are allocated to the destination node as an alternative of the source node. It computes those paths based cost and sends back to source node. B. Route utilization in PNDMSR The source node categorizes received three paths based on the cost and stores it in to the cache as primary path, secondary path and ternary path. The source node first selects primary path for data transmission to destination if it fails then next it selects secondary path for data transmission to destination if it also fails, then finally source selects ternary path for data transmission to

3 destination. If it also fails it invoke new discovery to find new energy aware node-disjoint path to destination. C. Route maintenance in PNDMSR The route maintenance in the PNDMSR is same the route maintenance in the DSR [9]. If currently using route is broken, then intermediate node sends a route error (REEP) packet to source node to inform the source node about the broken link. After receiving the REEP, the source node removes that broken route from its cache and it uses the alternative route to destination if it is available otherwise it invokes new route discovery to find new routes to destination. III. Performance Analysis of PNDMSR by using NS2. In this section, we describe the analysis and simulation setup using NS2 simulator [11].The NS2 is popular and standard simulation tool for implementing all types of network protocols, it was developed by the University of California at Berkeley and it is widely used by research scholars. The NS2 was developed by using two computer programming languages one is C++ another is OTcl (Object oriented Tool Command Language) interpreter as a command and configuration interface. The Network Animator (NAM) is an animation tool for inspecting network topology and data transmission shown fig.1.. Ubuntu operating system was used. The area for network configuration is chosen as 100m x 100m and the nodes are deployed within this region. The following quantitative performance metrics were used to measure the performance of both protocols Overhead Throughput Packet Delivery Ratio Packet loss Ratio End to End Delay of packets Energy Consumption By varying the following parameter Number of nodes (Density) [25, 50, 75, 100] Fixed {speed=2m/s, pause time=4ms, data rate=50kbps and load=4} Mobility of node(speed) [2 m/s, 4m/s, 6m/s, 8m/s] Fixed {network size=50, pause time=4ms, data rate=50kbps and load=6} Data Rate [25Kbps, 50Kbps, 70Kbps] Fixed {network size=50, speed= 2m/s, pause time=4ms and load=6} Pause time [2s, 4s, 6s, 8s] Fixed {network size=50, speed= 2m/s, data rate =50kbps and load=6} Number of source nodes and Destination pairs (increasing load) [2, 4, 6, 8] Fixed {network size=50, speed= 2m/s, pause time =4ms, data rate =50kbps} Table.1 shows different parameters used for simulation. A. Analysis of Routing Overhead The routing overhead is defined as the number of control packets are generated at the time of simulation. The overhead most important performances metric to compare the routing protocols Routing Overhead= Number of Control Packets (RREQ, RREP, REEP) generated by routing protocol during simulation. Fig. 1. Sample Mobile Ad Hoc Network shown in NAM (Network Animator with 50 nodes) Different performance metrics are employed to test the PNDMSR against the MSDR both are implemented using Network Simulator 2, in the new version is NS2.23,

4 Table.1.Simulation set up and parameters Sno a Parameter and Values 1 Network size and Node s Placement 100 x 100 Sq meters (Size of NAM), 2 Number of Nodes: 25, 50, 75, Total Simulation time : 300s 4 Node s Mobility : Random Way Point Mobility Model(RWPMM) 5 Node s Speed: 2 m/s, 4m/s, 6m/s, 8m/s 6 Node s Pause Time: 2 ms, 4ms, 6ms, 8ms 7 Application Layer : Constant Bit Rate(CBR) Packet Payload size: 512 bytes Data rate :25Kbps, 50Kbps,75Kbps,100Kbps Number of source and destination pairs : 1, 2, 4,8 8 Transport Layer: User Datagram Protocol ( UDP),,CBR 9 Network Layer: PNDMSR and MDSR Routing Protocols 10 Data Link Layer: Logical Link Control Layer (LLC) Medium Access Layer(MAC):IEEE DCF 11 Physical Layer: Antenna Model : Omni Directional Wave Propagation Model: Two-way Ground Model Channel Type: Wireless Channel Physical layer Channel Bandwidth: 1Mbps Interface Queue Type: CMU Priority Queue with size 50 packets Transmission Range of each node: 914 MHZ, Lucent Wave LAN card with 250 meters Interference range: 500 meters 12 Energy Model: Initial energy of each node: 100Joules Transmission Power: 0.7 Watts Receiving Power: 0.3Watts Idle Power: watts 13 RREQ_Wait_Time and RREP Wait_Time: 0.5ms The normalized routing overhead is calculated as follows: x Normalized Routing Overhead= (1) y Where x is total number of routing packets generated and y is total number of packets received 1) Overhead by Varying Number of nodes Fig.3.1. Overhead Vs Number of nodes Fig.3.1. shows, Overhead Vs Number of nodes (density) by varying number of nodes 25, 50, 75 and 100. The low overhead for both protocols is at 25 nodes, the highest overhead occurs at network size with 75 nodes. Usually the overhead is directly proportional to network size, but in this case overhead of both protocols has been increased in ascending order from 25 to 75 nodes. But suddenly overhead is drastically reduced at network size with 100 nodes due of dynamic characteristics of ad hoc network. However, the overhead is not purely depends on networks size. It also depends on unpredictable in node mobility, poor wireless link quality, error-prone radio shared channel, hidden and exposed problems. In all four cases PNDMSR produces less overhead than MDSR, because the PNDMSR prevents the overhead by eliminating excessive control packets (RREQ and RREP) using broad cast with low overhead approach and it also selects the energy aware paths that are least cost (higher energy).the main objective of the PNDMSR is selecting optimal energy aware node-disjoint multipath by optimizing the overhead using node s cost. 2) Overhead by varying node s speed Fig 3.2 shows the Overhead Vs Speed of node. Here node s speed is varied 2m/s, 4m/s,6m/c and 8m/s. Usually speed increases overhead is also increases because probability of path failure increases

5 4) Overhead Vs Varying Data Rate Fig.3.2 Overhead Vs Speed (m/s) At low mobility, the probability of path failure is low. At high mobility, selected path will be failed due to frequently links failure or node s mobility. The PNDMSR produces less overhead than MDSR and it performs much better than MDSR when speed increases. The PNDMSR prevents the overhead by eliminating excessive control packets (RREQ and RREP) during route discovery itself. The mobility of node increases the overhead of the MSDR is drastically increases compare to the PNDMSR. 3) Overhead Vs Varying pause time Fig Overhead Vs Data Rate (Kbps) Fig. 3.4 shows overhead Vs Data rate, here data rate is varied 25 Kbps, 50Kbps and 70Kbps and usually data rate is directly proportional to the overhead. When data rate is 50 Kbps and 70Kbps, the overhead of the PNDMSR is less than overhead of the MDSR. At 25Kbps the overhead of the PNDMSR has high overhead than MDSR, overhead is not purely depends on the data rate. Increase the data rate the PNDMSR yields less overhead than overhead of MDSR. 5) Overhead by varying Source and Destination pairs (connections)/load. Fig.3.5 shows overhead Vs number of sources and destination pairs (load), here number of sources is varied 1, 2, 4 and 8. The number of sources (load) is directly proportional to the overhead. When increasing load, The PNDMSR has less overhead than overhead of MDSR and the overhead of the MDSR drastically increases compare to the overhead PNDMSR Fig Overhead Vs Pause Time Fig.3.3 shows, Overhead Vs pause time, here pause time is varied 2ms, 4ms, 6ms and 8ms. Usually, the pause time increases network stability also increases. However, in the three cases at 2ms, 6ms and 8ms, the PNDMSR produces less overhead than the overhead of MDSR except in case of the pause time at 4ms, because due of dynamic characteristics of ad hoc network. The overhead is not purely depends on the pause time. However, it also depends on unpredictable dynamic characteristics of network, mobility of node, poor wireless link quality, error-prone radio shared channel, hidden and exposed problem. Fig.3.5. Overhead Vs Number of Source and Destination pairs (connections)

6 B. Throughput (Kbps) analysis: To measure the protocol performance, throughput serves as the better parameter. The throughput is defined as the ratio of number of packets received to the number of packets transmitted [10] and it is indirectly proportional to the overhead. The throughput is calculated by using the equation 2. x 8 Throughput= Kbps (2) t 100 Where x is number of bytes received and t is simulation time Fig 3.8.Throughput Vs Pause time The following five cases show that the PNDMSR yields higher throughput than the throughput of MDSR because overhead of PNDMSR is less than overhead of MDSR Fig Throughput Vs Data Rate (CBR) Fig Throughput Vs Number of Nodes (Network size/density) Fig Throughput Vs Number of Source and Destination pairs (connections) C. Analysis of Packet Delivery Ratio (PDR) Fig 3.7. Throughput Vs Speed To find the efficiency of the protocols, PDR is one of the important qualitative metrics. It is defined as the ratio of data packets received and packet sent [10], it is calculate as follows

7 x PD R = 100 (3) y Where x is the total number of packets received and y is the total number of packets sent at end of the simulation time. The overhead is indirectly proportional to packet delivery ratio. The following all five cases show that overall the PNDMSR yields higher PDR than the PDR of the MDSR because overhead of PNDMSR is less than the MDSR Fig PDR Vs Pause time Fig PDR vs Number of Nodes Fig PDR Vs Data Rate (load) Fig PDR Vs Speed Fig PDR Vs Number of Source to Destination Pairs D. Analysis of Average End-to-End Data Packet Delay (ms) To measure the performance of the proposed protocols end to end delay is the major performance factor which should be considered. The Average end to end delay is defined as the time taken to transmit the packet from source node to the destination node. The total end to end

8 average delay of data packets includes queuing delay, buffering delay, propagation delay and retransmission delay. The average end to end delay of n number of received data packets [10] is calculated by using the n ( R i S i) following equation 4. i = 1 delay = (4) n Where R i denotes the time at which the packet i is received and S i denotes the time at which the packet i has been sent to the destination. or it also defined as follows Total delay of each data packets Average delay = Total data packets received. Fig Delay Vs Pause Time The overhead is directly proportional to the packet delay, The following five cases show that overall, the PNDMSR yields less average end-to-end packet delay than the average end-to-end packet delay of MDSR except in case of delay Vs pause time is show in fig 3.18, pause time at pause time at 2 and 8 ms, the PNDMSR has high average end to end delay compared to average end to end packet delay of MDSR. Fig Delay Vs Data Rate (Kbps) Fig Delay Vs Number of Nodes Fig Delay Vs Number of Source and Destination pairs E. Packet Loss Ratio (PLR) (%) Fig Delay Vs Speed The PLR is used to measure the performance of the routing protocols. The PLR is defined as the ratio of number of packets not delivered to the destination form the total number of packets sent [10]. The PLR is calculated by using the equation

9 x y PLR = 100 (5) x Where x is total number of data packets sent and y is total number of data packets received. Number of data packets lost=x- y. Some data packets not reached to destination and they are lost during transmission to due congestion and collision in the network.the overhead is directly proportional to congestion, probability of collisions, packet loss, packet error rate. The following five cases show that overall the PNDMSR has less packet loss compared to the packet loss of MDSR because the PNDMSR has yielded less overhead compared to the overhead MDSR. Fig Packet loss Vs Data Rate (Kbps) Fig Packet loss Vs Number of Source and Fig Packet loss Vs Number of nodes F. Energy Efficiency It is an important qualitative metrics to measure the performance of different protocols. The energy efficiency is calculated as the ratio of total received data (bytes) to the total energy consumption using equation 6[10] x Energy Efficiency= (6) y Where x is total received data in terms of bytes and y is total energy consumption in terms of joules Fig Packet loss Vs Speed (m/s) Total energy consumption= ( E R ) (7) n i= 1 i i Where E denotes the initial energy of i th node, i denotes the residual energy of i th node and n total number of nodes in the network. R i Fig Packet loss Vs Pause Time (Sec) The overhead is directly proportional to the energy consumption. The following four scenarios show that PNDMSR has less energy consumption compared to the energy consumption of MDSR because the PNDMSR has yielded less overhead compared to overhead the MDSR

10 Fig Scenario 1: Energy Consumption Vs Simulation time (Network Size=50) Fig Scenario 4: Energy Consumption Vs Simulation time (Network Size=50) IV. CONCLUSIONS Fig Scenario 2: Energy Consumption Vs Simulation time (Network Size=50) In this paper, we made a contribution to measure the performance of two different multi-path dynamic routing protocols such as PNDMSR and MDSR. The performance of the two protocols is tested with well-known simulator called as NS2 [11]. The performance of the PNDMSR and MDSR are tested by using various quantitative performance metrics like, end-to end delay of packets, throughput, routing control overhead, packet delivery ratio, energy efficiency, and packet loss rate and by varying various parameters like network s size, mobility of node, pause time, data rate and load.it is proved that PNDMSR has better performance than MDSR and it is concluded that the PNDMSR is one of scalable, robust and energy efficient multipath routing protocol in MANET. REFERENCES Fig Scenario 3: Energy Consumption Vs Simulation time (Network Size=50) [1] C.S.R. Murthy and B.S.Manoj, Ad hoc Wireless Networks: Architectures and Protocols, Prentice Hall, [2]M. Bheemalingaiah, C. Venkataiah, K. Vinay Kumar, M. M. Naidu, D. SreenivasaRao, Survey of Energy Aware On-demand Multipath Routing Protocols in Mobile Ad Hoc Networks, International Journal of Advanced Research in omputer Science and Software Engineering,Volume 6, Issue 4, April [3] YibinLiang, thesis on Multipath Fresnel Zone Routing for Wireless Ad Hoc Networks, Virginia Polytechnic Instituteand State University, March, [4] Zhenqiang Y, Krishnamurthy S. V and Tripathi S. K. A Framework for Reliable Routing in Mobile Ad hoc Networks, InProceedings of IEEE INFOCOM, 2003, Vol. 1, pp [5]Mahesh K. M and Samir R. D, On-demand Multipath Distance Vector Routing in Ad hoc Networks, In

11 Proceedings of IEEE International Conference on Network Protocols, pp.14-23, [6] S. J. Lee and M. Gerla, Split Multipath Routing with Maximally Disjoint Paths in Ad hoc Networks, InProceedings of IEEE ICC, pp , [7]L. Wang, Y. Shu, Z. Zhao, L. Zhang and O. Yang, Load Balancing of Multipath Source Routing in Ad hoc Networks, InProceedings of IEEE ICCC, 2002, Vol. 5, pp [8]M. Bheemalingaiah, M. M. Naidu, D. SreenivasaRao and G. Varaprasad, Power Aware Node- Disjoint Multipath Source Routing in Mobile Ad Hoc Networks,International Journal of Mobile Network Design and Innovation, Vol. X, No. Y, [9] J. Broch, D. Johnson, and D. Maltz, Dynamic Source Routing Protocol for Mobile Ad Hoc Networks, IETF Internet draft, [10]P.Periyasamy and. E.Karthikeyan, Energy Optimized Ad hoc on-demand Multipath Routing Protocol for Mobile Ad hoc Networks, International Journal of Intelligent Systems and Applications, vol11, pp: 36-41, [11]Network Simulation 2 official website: