Interface Assignment, Bandwidth Allocation And Scheduling For Wireless Mesh Networks

Recent years,Wireless Mesh Networks(WMNs) have emerged as a promising technology that provides wireless broadband accessibility and thus,extends the Internet connectivity at the edge and improves the network coverage and economy efficiency.Using WMN as a backbone for large wireless access imposes high bandwidth requirements.At the same time,the interference among simultaneous transmissions may dramatically cause capacity reduction.Fortunately,Multiple Channel Multiple Interface(MCMI) and Multiple Input Multiple Output(MIMO) are two of the new technologies introduced to mitigate interference and increase simultaneously transmissions.The first solution is to use multiple non-overlapping channels on the interfering links to maximize the parallel transmission and throughput.For example,IEEE 802.11b/g and 802.11a standards specify 3 and 12 non-interfering channels,respectively.The ability of using MIMO antenna technology in the wireless transmission is another major technique to alleviate the problem,and thus improve throughput substantially.In this dissertation,our research focuses on interface assignment,bandwidth allocation and scheduling for WMNs using both technologies,respectively:●With the ability of simultaneous transmissions,MCMI WMNs have emerged with great potential in the improvement of network throughput and fairness. However,most proposed channel assignment algorithms for WMNs made an assumption that the Network Interface Cards(NICs) are evenly assigned to the mesh touters.In Chapter 3,we investigate the problems of NIC assignment and bandwidth allocation to minimize the infrastructure cost, and meanwhile guarantee the application requirements.We argue evenly assigning NICs to all routers is neither a necessary condition nor an effective solution,since not only the interference but also the traffic flows are important factors that will affect the parallel use of the bandwidth.One of the principal challenges addressing these problems is their interactive impact on the optimization of network throughput.By analyzing all kinds of constraints for traffic,NIC and performance,we formally define a problem space that addresses the relationships between different assignment and allocation problems.Furthermore,we demonstrate that a hard NIC assignment and bandwidth allocation problem can be decomposed and formulated into a well-defined single or multiple-phase problem.In addition to the Linear Programming(LP) solutions,we propose novel efficient heuristics for on-line decisions for the situation whenever the network architecture changes and needs recompute the system performance in real-time.We show through extensive simulations that the heuristic algorithms can achieve close to optimal solution and outperform the equal NIC assignment method with even a smaller number of NICs.●Over conventional antenna technologies,MIMO technique have presented great ability in the improvement of network capacity with the unique features of spatial multiplexing and interference suppression.In order to exploit the benefit of simultaneous transmissions provided by MIMO,researchers have proposed a number of cross-layer optimizations and MAC layer designs to increase the throughput of wireless mesh networks,where the number of elements in the antenna arrays are pre-allocated or evenly assigned to the routers.In Chapter 4,we argue that using the same number of elements in each antenna array in all routers is not a necessary condition for the improvement of system performance.This is because the requirement for the number of elements is quite different for each router.Especially at those critical touters that have huge aggregate traffic toward the gateway,more elements are needed not only for the traffic relay but also for the interference suppression.Based on this observation,we define the joint problem of element assignment,bandwidth allocation and scheduling to characterize the throughput benefits of cross-layer optimizations.We propose a Cost-Aware Element Assignment(CAEA) technique to minimize the total number of the antenna elements when still achieving the optimal bandwidth allocation.In addition,to verify the efficiency of the CAEA assignment,a heuristic Trafficaware Stream-controlled Link Scheduling(TSLS) algorithm is proposed to provide a schedulable bandwidth allocation.We demonstrate through extensive simulations that our solutions(CAEA,TSLS) not only effectively save the total cost on antenna elements but also perform close to optimal.