Efficient flow-based channel assignment schemes for congestion avoidance in wireless mesh networks

The demand of fast and large bandwidth Wireless Mesh Network (WMN) to provide last mile Internet access has motivated high interest in enhancingWMN with multi-radio interfaces, where each radio is dedicated to a non-overlapping channel. However, as a result of contention-based nature of IEEE 802.11-based wireless mesh network, the node capacity is limited by the contention, interference caused by hidden nodes, data transmission over low rate links, and frequent interface switching. When a mesh node cannot win a sufficient number of transmission opportunities to satisfy its traffic load, it becomes saturated. In this case, more data packets are stored in the queue and congestion is occurred. Additionally, with the spanning tree structure of gateway traffic, where most of the traffic between gateway and mesh nodes, nodes near the gateway become congested. This thesis presents efficient flow-based channel assignment schemes for congestion avoidance in WMN. The most significant contribution of this thesis is to design on-demand channel assignment that helps in avoiding node/link congestion by assigning non-overlapping channels having less interference to every link on the established path and avoiding the channel/link congestion at the critical links in WMN. As a result, this thesis proposes a centralized on-demand channel reservation scheme (AODV-MRCR) that provides an efficient way of utilizing the multiradio and multi-channel resources to establish high throughput path for the gateway traffic. To ensure high throughput path, AODV-MRCR uses path optimization as an intelligent mechanism to select a path with least interference and eliminate the intra-flow interference. In addition, the scheme develops multilink routing discovery process to solve the problem of single routing entry of the multi-source single destination flow traffic. In the proposed scheme, the channelto- interface binding is integrated with reactive gateway discovery process and a hybrid cross-layer mechanism is developed for the channel negotiation and synchronization. Next, AODV-MRCR is extended to address the problems of throughput limitation and reduction which are caused by channel assignment overhead, interference caused by hidden nodes, and link congestion at the critical links in WMN. The result is AODV-CSHDIA, distributed on-demand carrier sense and hidden node channel assignment scheme. Therefore, a channel selection metric is developed based on analytic throughput model with existing hidden nodes. This helps to avoid the collision caused by the on-going transmission of the hidden nodes. So, the network throughput will be improved. In addition, a hybrid interface assignment strategy is developed based on the spanning tree structure of the gateway traffic. This minimizes the interface switching and channel negotiation overhead of the hybrid channel assignment. The proposed channel assignment is integrated with the reactive and proactive discovery process to establish high throughput paths for local and gateway traffic. A receive-forward algorithm is developed to integrate the channel assignment with the proactive routing discovery process to assign channels for the links of the gateway traffic. Finally, flow-based channel assignment is developed for rate separation in multirate, multichannel and multi-radio WMN. AODV-RDCA is introduced to avoid the capacity reduction of high data rate links caused by low rate links when they share the same channel. For this, a novel route metric is designed with the purpose of selecting a path with a minimum number of low rate links and less interfering with existing flows. The metric is developed based on the actual channel throughput. Also, two algorithms are developed to address the link sharing problem during the channel assignment stages; channel-to-interface and link-to-interface stages. To achieve that, system throughput channel selection metric is developed considering the effect of low rate link at both carrier sense and hidden ranges on the channel throughput. This metric is developed based on analytic throughput model taking into account the impact of low rate links and the hidden nodes on the channel throughput. Several simulation scenarios and analytical model have been presented in order to evaluate and compare the proposed schemes with the existing channel assignment. The results demonstrate that the proposed schemes utilize WMN characteristics to use the spectrum of the resources efficiently and improve the WMN throughput significantly