Research On Transmission Performance Of Wireless Ad Hoc Networks Based On Stochastic Geometry Theory

With the rapid evolution of wireless communication technology, the network size and appliedrange enlarge dramatically, the diversity of node types and network forms expand significantly andlead to increasingly higher complexity in Ad Hoc networks. Therefore, to guarantee the optimalsetup of a network it’s mandatory to evaluate the metric for some specific network form and tocompile the performance evaluation before. However, due to missing centralized control anduncoordinated transmission, unexpected interference between the communication nodes occurfrequently. In such cases, the performance-limiting metric is the signal-to-interference-plus-noise-ratio (SINR) rather than the signal-to-noise-ratio (SNR). Additionally, owing to the dynamictopology and the interference between multiple senders and receivers which increases thecomplexity between sender and receiver systems, the links cannot be determined. Thus, the classicShannon information theory based on point-to-point systems can not be used to analyze these newcharacteristics of Ad Hoc networks. To explore a new and rationally analytical framework, whichsupports effective performance evaluation of Ad Hoc networks, is extremely complex, but urgentlyneeded. This dissertation focuses on performance evaluation in four possible network scenarios inthe future. The analytical framework is based on stochastic geometry, aiming to draw up anSINR/SIR analysis model to derive closed-form expressions for the performance metrics, such asthroughput, outage probability, success probability, network capacity, etc. on which then differentcommunication strategies could be studied and the intrinsic relationship between the transmissioncapability and network parameters could be investigated, which finally lay a solid foundation forthe network planning and optimization. The main achievements are as follows:Firstly, a scenario is considered in which multiple cognitive radio Ad Hoc networks arecoexisting and sharing an unlicensed band. Based on the shot-noise model from stochastic geometry,the characteristic function of the interference in each network (not only from its own network butalso from the other networks) is detected, which then allows to derive the closed-form expressionsof the per-node throughput of each network in AWGN channel and Rayleigh fading channel. Next,in order to mitigate the level of interference, the assumption is that each receiver has a protectionzone, and no other nodes within this zone have the permission to transmit. In this case, theinterference for each network is modeled as a Gaussian random variable. Utilizing stochasticgeometry, its mean and variance is applied by AWGN channel, Rayleigh channel and Nakagami-m fading channel respectively, and so enables the determination of per-node throughput of eachnetwork. Among different Ad Hoc networks, transmission probabilities could be coordinated bycentralized control, where transmission control criteria are used to optimize the throughputperformance. Simulations verified the theoretical findings of the analyses and results aredemonstrating that there is a tradeoff between high utilization and fairness when proportionalfairness criterion is used.Then, we investigate the success transmission probability of the authorized users in a cognitivecooperative network. In an overlaid cognitive radio network, it is inevitable to cause interference tothe authorized user due to inaccurate spectrum detecting by the cognitive user. When the authorizeduser fails to decode its information signal, one of the cognitive transmitters of correct detection willbe selected to serve as a sole relay to help forwarding. In this dissertation two cooperativetransmission schemes are analyzed, the distance based and the Signal-to-Noise-Ratio (SNR) basedschemes. Assuming that the cognitive user transmitter location is random, particularly it is modeledas a HPPP. Utilizing stochastic geometry, the closed-form expressions of success probabilities withthose two cooperative transmission schemes in authorized systems are derived. Simulation resultsconfirm the analytical derivation and indicate that the SNR based scheme has the best performanceand the distance based scheme is superior to the PU direct re-transmission scheme.Next, a new framework is formulated to investigate the capacity and end-to-end delay in themulti-hop Ad Hoc network where each node is equipped with multi-antennas. HPPP is introduced tomodel the locations of nodes over the plane whereas TASEP is used to model the transport ofpackets for a particular flow/route. The metrics to characterize the multi-hop Ad Hoc networks aredefined as the spatial capacity density and the average end-to-end delay. The closed-formexpressions of these two main end-to-end performance metrics are derived respectively under threehopping strategies-closest neighbor, furthest neighbor and randomly selected neighbor hoppingstrategies. Based on numerical simulation, the relations among node density, diversity gains,number of relays and related network parameters, which jointly determine network performance,are investigated and these guidelines provides insights in how to select the best hopping strategyfrom a real network design perspective.Finally, we study the transmission capacity under a constraint of maximum packet errorprobability in a Ad Hoc network when real modulation and coding schemes are used. Based ondifferent modulation/coding schemes and assuming channel inversion power control strategy, outage probability and transmission capacity of networks are investigated under a constraint ofmaximum packet error probability. The numerical results are showing that the maximumtransmission capacity is not only related to the choices of the modulation and coding schemes, butalso the transmission density. Based on the given transmission density in a real network, it isnecessary to make appropriate selection of modulation and coding schemes in order to maximizethe transmission capacity.