Wireless mesh networks are a profitable application of ad-hoc networks for their low installation cost and high performance properties. Here, we are concerned with the problem of how the mesh nodes should be placed in a given service area to satisfy some network constraints. We investigate this problem by means of genetic algorithms, which are used in order to find optimal topologies of a mesh network. The resulting algorithm has been called as GAMesh, and it is a numerical framework where the design constraints of the network, such as the connectivity and the traffic constraints, can be easily written by means of a scalar fitness function. We cast the problem in an innovative graph drawing technique, which furnishes good solutions in a reasonable computational time. Differently from other solutions, GAMesh explores all the points in the discretized plane. We show GAMesh in the context of infrastructure based wireless mesh networks, typically used by wireless internet service providers, which must minimize cost and maximize the performance.
展开▼
