Gateway selection optimization with QoS constraints in static multi-hop wireless networks

Abstract

In a static multi-hop wireless network, some nodes are chosen as gateways to the Internet while the other nodes cooperatively route traffic in a multi-hop fashion to the Internet through the gateways. Most traffic is directed to and from gateways for Internet access and thus the placement of gateways is critical to the performance and management of the network. A good gateway placement should provide throughput/delay guarantees to users and use wireless bandwidth efficiently. In this thesis, we propose an approach to optimize the gateway selection with QoS requirements such as throughput and delay. The optimization objectives are selecting a minimum number of gateways and balancing the load of gateways while satisfying QoS requirements. We employ collision domains, the sets of contending links sharing the capacity of the same channel, to model accurately the impact of wireless interference on network throughput. Accurate interference estimation increases the efficient use of wireless bandwidth by limiting the number of flows sharing the same channel, and thus it helps ensure throughput requirements to users. Balancing the load of gateways helps reduce contention at heavily-loaded gateways, and hence it increases the network resource utilization efficiency. We formulate the selection problem as a Multi-Objective Linear Program and design an efficient greedy algorithm for it. We extensively evaluate our approach and show that: (1) it satisfies node throughput requirements to a greater extent than do previous approaches; (2) it benefits not only single-radio single-channel networks but also multi-radio multi-channel networks; and (3) the greedy algorithm yields good suboptimal solutions.