Performance Optimization For Wireless Ad Hoc Networks Based On Interference Management

Without the assistance of any infrastructure, distributed wireless users can quickly formwireless ad hoc network (WANETs) on-line, so as to support various applications. Such kindof networks are known to be of low cost, high self-healing ability and strong scalability. Sincetheir first applications in the military field, they have now been widely applied in many civilfields, e.g., wireless sensor networks, wireless mesh networks, etc.Due to the open and shared nature of wireless medium, concurrent transmitting links sufferinevitable interference from each other. Thus, how to manage such mutual interference towardsmaximizing the network performance has been a longstanding research focus. Traditional tech-niques of interference management, such as link scheduling, power control, rate control andtopology control all follow the interference avoidance approach. On the contrary, recently pro-posed interference management techniques from the physical layer, e.g., interference neutral-ization, interference cancellation and interference alignment, directly remove the interferencesignal at the receiver. Concentrating on these traditional and new interference managementtechniques, research in this dissertation is carried out along the following four directions.1) EnergyefficiencyandthroughputaretwoimportantperformancemetricsforWANETs,which decide the network lifetime and service quality respectively. In multi-rate802.11WANETs, energy efficiencies as well as minimum required transmission power of differ-ent rate level differ a lot from each other. Hence, this dissertation proposes a traffic-awareactive link rate adaptation algorithm to optimize the network performance. By constant-ly monitoring the length of interface queue for each link, the algorithm can tell whetherthe current link rate can satisfy the traffic demand or not, hence selects the right link rateas well as the right transmission power, which eventually increases the network energyefficiency as well as network throughput substantially.2) The multi-hop characteristic of WANETs requires end-to-end (e2e) transmission rateprovision for each user. Random competitive link scheduling can not guarantee the chan-nel access time of each link, thus fails in providing accurate e2e transmission rate. Hence,this dissertation proposes a cross-layer solution based on cooperative link scheduling un-derthephysicalinterferencemodel, whichjointlyconsidersroutingandadmissioncontrol to guarantee e2e transmission rate for each user. The link scheduling module allows eachlink reserves their transmission slots in a distributed way, while guaranteeing that linktransmission will not be failed by interference during the reserved slots. The routing mod-ule chooses the best path to maximize the performance of link scheduling performance,i.e., minimizing the number of required slots for a given e2e transmission rate require-ment. When the traffic load exceeds the network capacity, the admission control modulewill reject new users so as to guarantee e2e rate for existing users.3) Network capacity is of vital importance for WANETs, while traditional topology con-trol based on interference avoidance neglects the potential capacity enhancement broughtby the newly proposed interference neutralization (IN) technique. Hence, this dissertationstudies how to further boost network capacity for single-antenna WANETs via topologycontrol incorporating IN. Specifically, this dissertation abstracts IN between neighboringlinks as a new cooperative transmission pattern. Then by jointly considering this newtransmission as well as the existing non-cooperative transmission pattern and traditionalcooperative transmission pattern, this dissertation abstracts the topology control probleminto a transmission pattern optimization problem. After deriving the relationship betweentransmission pattern and network capacity, this dissertation proposes a distributed topol-ogy control scheme towards improved network capacity.4) Traditionallinkschedulingalgorithmensuresthequalityofeachlinkbyseparatingcon-current links either in the spatial domain or in the time domain. Only links far away fromeach other can possibly concurrently access the wireless medium. However, with the re-cently proposed technique of interference alignment (IA), concurrent transmission amongneighboring links are now possible. Hence, this dissertation revisits the minimum lengthscheduling problem in multiple-antenna WANETs, and propose a practical distributedlink scheduling algorithm. The proposed algorithm can support concurrent transmissionamong neighboring links by fully exploiting the IA technique, which in turn boost thescheduling performance.