Theses and Dissertations
Permanent URI for this collectionhttp://ir.daiict.ac.in/handle/123456789/1
Browse
6 results
Search Results
Item Open Access Game theory based strategies for cooperation in ad hoc wireless networks(Dhirubhai Ambani Institute of Information and Communication Technology, 2010) Rajput, Nitin Singh; Srivastava, Sanjay; Divakaran, SrikrishnanIn self-organized ad-hoc wireless networks, nodes belong to different authorities, pursue different goals, have constrains like energy; therefore, cooperation among them cannot be taken for granted. Non cooperation of nodes causes increase in probability of packet drop and increase in probability of route or network failure, which leads to poor network performance. On other hand if nodes always cooperate, network does not last for a long as nodes are energy constrained. Several schemes are proposed in literature based on incentive and reputation mechanism. All of them outperforms in their own set of assumptions and have certain issues.Researchers started looking at game theory as a probable solution and proposed some schemes, but yet to come up with better solutions. We focus on optimization of service received by a node from network and delivered to network considering energy as constraint. We first derive the probability by which a node accept the relay request of other nodes based on energy constraint. Then apply game theory based schemes Generous Tit For Tat (GTFT), Neighboring GTFT (N-GTFT) and contrite Tit For Tat (C-TFT) for acceptance of relay requests; well known in economics, behavioral science and biology for cooperation. We find that all above mention scheme converges towards parato optimal values of service received and deliver to network in presence and absentia of noise in network. We find that when network has pair wise mixed strategies (any two from GTFT, N-GTFT and C-TFT) then also convergences remain same as it is for single strategy. But when there are 50% non cooperative nodes (Always Drops strategy), C-TFT is the dominating strategy. Also C-TFT copes up with 10% or more noise in the network as other strategy fail to do so, however they cope well with lesser amount of noise. At the end CTFT evolve as the dominating strategy when all strategies including Always Drops simulated under evolutionary method of comparison in noisy and noise free network. C-TFT out performs because it switches to mutual Tit For Tat after cooperating against fixed unilateral non cooperation from other nodes and also cope up with own unintentional defection caused by noise.
Item Open Access Cooperation enforcement mechanisms in wireless adhoc networks(Dhirubhai Ambani Institute of Information and Communication Technology, 2009) Chaturvedi, Manish Shivshankar; Srivastava, SanjayWireless adhoc networks are autonomous, infrastrutureless networks where there are no dedicated routers or base stations and nodes are expected to cooperate in Wireless performing routing duties to keep network connected. Cooperation can be assumed if all nodes belong to single authority (e.g. military service or disaster management). But in applications where nodes do not belong to single authority and have limited resources (energy of battery driven devices), like pervasive computing or ubiquitous computing environment, one can not deny possibility of node selfishness. Also as there is no central authority to control node behavior, one can not deny possibility of node maliciousness. Many cooperation enforcement schemes are proposed in literature, and every scheme is shown to perform better under its own set of assumptions. These assumptions are different for different schemes. So, we aim at defining common set of assumptions and comparing selected schemes on this common ground with respect to packet delivery ratio, energy consumption, routing and other control overhead. We find that with the traffic scenario where nodes do not need services of one another at the same time, the cooperation schemes are not effective in dealing with node selfish ness. We also find that, while dealing with malicious nodes, all cooperation schemes perform better than Dynamic Source Routing(DSR) protocol in improving packet de- livery ratio(PDR), but this improvement comes at the cost of significant increase in routing control packets overhead and energy consumption of cooperation enforcement schemes is higher than that of DSR. Also they fail in punishing misbehaving nodes and the PDR of malicious nodes is comparable to that of good nodes.Item Open Access Buffer aware routing in interplanetary ad hoc network(Dhirubhai Ambani Institute of Information and Communication Technology, 2007) Mistry, Kamal; Srivastava, Sanjay; Lenin, R. B.Recent research has shown that existing TCP/IP protocol suit exhibits poor performance in space communication networks. To deal with communication challenges in deep space such as long propagation delay, high channel error rate, limited bandwidth etc. the architecture called Interplanetary Ad hoc Network (IPAN) is envisioned to establish a communication infrastructure in environment which allows to connect planets, natural and artificial satellites, and various mission elements such as space- crafts and rovers. The communication links in IPANs possess the properties of low bandwidth, high error rate, high latency and link unavailability for longer period of time. Nodes in such networks are resource constrained in terms of storage, energy and processing power. This work proposes a probabilistic routing protocol called \Buffer Aware Routing Protocol in Interplanetary Ad hoc Network (BARPIN)" based on the store and forward principle of Delay Tolerant Networking (DTN). Here IPAN is modeled as the network of two kinds of nodes, one having deterministic mobility patterns with greater resources and the others having random movement patterns with limited resources in terms of storage and energy. The delivery ratio performance measure of BARPIN is studied for different network conditions by changing field sizes, traffic rates, available resources on various nodes and for different source-destination pairs. Further we propose a model to estimate the minimum required buffer size of nodes for different data rates to reduce the packet loss due to buffer overflow, and justified the simulation results with analytical results.Item Open Access Modeling MANETs using queuing networks(Dhirubhai Ambani Institute of Information and Communication Technology, 2007) Munjal, Aarti; Srivastava, Sanjay; Lenin, R. B.Mobile Ad Hoc Networks (MANETs) are becoming an attractive solution to the services that require flexible establishment, dynamic and low cost wireless connectivity. Since nodes are mobile, routing results vary significantly with the underlying mobility model. So, modeling (nodes as well as wireless links between them) plays a critical role in the performance analysis of MANETs. This work involves modeling MANETs in two ways. In the first model, nodes are modeled as static and wireless links between a pair of nodes are available or unavailable for exponential durations. When the link is available, nodes are connected and can have data transmission and data are lost if the link is not available as the nodes are not connected. Second model involves mobility being captured by making the servers go ON/OFF for exponential amount of time, and no departure takes place while server is on vacation. This way one does not lose packets but the queuing delay increases. In this work, these two queuing networks are proposed to study to performance measures of MANETs. Numerical results are derived using mathematical equations and then verified through simulation.Item Open Access Mobility aware MANET routing protocol using cross layer design(Dhirubhai Ambani Institute of Information and Communication Technology, 2006) Patel, Komalben; Srivastava, Sanjay; Lenin, R. B.Mobile Adhoc Network (MANET) is a dynamic network with time varying topology and time varying network resources. Due to the error-prone wireless channel and high mobility, traditional protocols of wired networks cannot be successfully applied to MANETs. The popularity of mobile and hand held devices equipped with wireless interface is creating a new challenge for Quality of Service. The wired network has also not been able to fulfill end-to-end guarantees. Due to the nature of MANETs, achieving the same end-to-end guarantees is very difficult. The mobility rate makes the task dicult. The aim is to fight against the losses caused due to mobility. This work attempts to build stable paths so as to counter the effects of mobility induced route failures. Using the Cross Layer Approach, the signal strength of the link from the MAC Layer is captured and used at the network layer. Upon receiving the signal strength values, the network layer carries out a prediction mechanism to predict the future signal strength. This information is then used to categorize the link as stable or unstable. The work also deals with designing a proactive routing protocol which uses the information of stable and unstable links to build up routes using limited dissemination technique. The protocol is similar to distance vector protocol in which, only the distance vectors whose next hop is a stable is advertised. This ensures that only stable link information reaches the other nodes in the network. The protocol has been tested in Network Simulator-2 and compared with the Destination Sequenced Distance Vector [PB94] and the Adhoc On-Demand Distance Vector [PR97] protocols. Various parameters like percentage of packet delivery, packet latency, effects of moblity, etc. are measured. The results show that the performance of the designed protocol is better than AODV in high traffic scenarios. With less overhead, the proposed protocol always performs better than DSDV. In high mobility scenarios, the protocol is comparable to AODV and performs better than DSDV.Item Open Access Selfish routing and network creation games(Dhirubhai Ambani Institute of Information and Communication Technology, 2005) Gandhi, Ratnik; Chatterji, SamareshThis work studies the two important problems of routing and network creation in the situation of selfish behavior of agents. In routing, agents want to send their data from source to destination. They try to reduce cost incurred in the process of routing. In network creation, agents create agent-to-agent link to form a network on which they can Communicate. Here there are two types of cost incurred: link creation cost and routing cost. Each agent tries to reduce his own cost. To study degradation caused by selfish behavior of agents we primarily use the standard notation of Price of Anarchy, which is ratio of the cost incurred at Nash equilibrium to the optimal cost. We show some results on Price of Anarchy and on different cost functions for above two problems, we propose a new model in network creation and show a polynomial time algorithm to verify Nash equilibrium.