Study On The Impact Of Users' Selfish Behaviors On Network Performance

A computer network is a collection of computers and other related devices which are connected to each other.The network allows computers to communicate with each other and share resources and information.On the one hand,the performance of networks can determine the user behaviors.For example,users will decide whether to join a network according to the network performance;On the other hand,due to developments of computer technologies and network technologies,the user behaviors have more and more significant impacts on the performance of networks.This thesis studies the user behaviors in both wireless networks and peer-to-peer(P2P) networks,and their impacts on the network performance.The main results of the thesis are as follows:1.Selfish power control in wireless local area networks:This thesis studies the selfish power control behavior and its impact on the performance of a wireless local area network(WLAN),where each link is under a throughput requirement.Employing the modulation scheme Non-coherent Frequency Shift Keying(NFSK),it is proved that there is an optimal transmission power and a corresponding optimal transmission strategy for a link with fixed channel condition.A repeated dynamic game is proposed,assuming that there is only one link which can adjust its transmission strategy in each unit time and the link always adopts the self-optimal strategy.Since links are motivated to move transmission power from high interference durations to low interference ones,this game can not converge in some cases.A penalty function is introduced to ensure convergence and the performance of various penalty functions is investigated.Simulation results show that heavier penalty leads to faster convergence,but has no direct effect on the energy consumption of the steady state.2.Selfish rate adaptation in wireless multi-hop networks:This thesis studies the selfish rate adaptation behavior and its impact on the performance of a wireless multi-hop network,where each link is under a throughput requirement.In an IEEE 802.11 wireless multi-hop network,it is proved that there is an optimal transmission strategy for a link with fixed channel condition.A repeated dynamic game is proposed,assuming that there is only one link which can adjust its transmission strategy in each unit time and the link always adopts the self-optimal strategy.Due to the selfish rate adaptation behavior,the link scheduling order affects the feasibility of the throughput requirement as well as the total power consumption.A pricing function is introduced to motivate links to share the channel fairly and efficiently.Simulation results show that the proposed approach leads to not only more feasible solutions,but also less total power consumption.3.A stochastic model for BitTorrent systems:In BitTorrent systems,the arrival and departure of peers lead to peer dynamics;peers choose neighbors according to the mechanism optimistic unchoking and the principle tit-for-tat,which incurs connection dynamics.This thesis studies the peer dynamics and the connection dynamics,and it is observed that peers with different connection numbers behave differently while choosing neighbors.So we define peer state,and system state which is composed of the peer states for all the peers in the system.We compute the transition rates between different system states when dynamics happen,and then get a stochastic model based on Markov chain. Using this model,the system performance at the steady state is studied.The accuracy of the model is verified by simulation,and it is shown that the file size has significant influence on the average number of connections per peer at the steady state.