No abstract

Efficient broadcast schemes are essential in wireless mesh networks (WMNs) for minimizing the content update time. In this paper, we consider the widest spanning tree problem in a multi-channel multi-interface WMN, where the width of a tree is determined by the bottleneck link bandwidth. To the best of our knowledge, we present the first effort in solving the widest spanning tree problem using mathematical formulation. In our model, we jointly consider and solve the problems of channel assignment, routing, scheduling and server/root placement. Unlike other spanning tree approaches, we allow WMN nodes to have heterogeneous number of network interface cards (NICs), and multiple NICs of a node can share the same assigned set of channels. To find a practical schedule, we also introduce the channel conflict graph and NIC constraint graph, and show that the associated scheduling problem is equivalent to the classic graph coloring problem.

We consider multi-channel multi-interface wireless mesh networks with a schedule-based MAC protocol, where conflict-free transmission is ensured by requiring links assigned with the same channel and within the mutual interference range of each other to be active at different time slots. When a (point-tomultipoint) multicast call arrives, the call is accepted if a multicast distribution tree can be established for connecting the source node with all the receiving nodes, and with sufficient bandwidth reserved on each link. Otherwise, the call is rejected. To maximize the call acceptance rate, the multicast tree must be constructed judiciously upon each call arrival. Aiming at minimizing the carried load on the most-heavily loaded channel, and maximizing the residual capacity of the most heavily loaded node, an integer linear program (ILP) is formulated for multicast tree construction. Since solving ILP can be time-consuming, an efficient heuristic algorithm is then proposed. We compare the two tree construction algorithms by simulations. We found that both algorithms give comparable call acceptance rate, but the heuristic algorithm requires much shorter running time.

In this paper, we consider the problem of constructing bandwidth-guaranteed multicast tree in multi-channel multiinterface wireless mesh networks. We focus on the scenario of dynamic multicast call arrival, where each call has a specific bandwidth requirement. A call is accepted if a multicast tree with sufficient bandwidth on each link can be constructed. Intuitively, if the carried load on both the most-heavily loaded channel and the most-heavily loaded node is minimized, the traffic load in the network will be balanced. If the network load is balanced, more room will be available for accommodating future calls. This would maximize the call acceptance rate in the network. With the above notion of load balancing in mind, an Integer Linear Programming (ILP) formulation is formulated for constructing bandwidth-guaranteed tree. We show that the above problem is NP-hard, and an efficient heuristic algorithm called Largest Coverage Shortest-Path First (LC-SPF) is devised. Simulation results show that LC-SPF yields comparable call acceptance rate as the ILP formulation, but with much shorter running time.

No abstract