We consider the problem of selecting a subset of p out of n sensors for the purpose of event detection,in a wirelesssensor network（WSN）.Occurrence of the event of interest is modeled as a binary Gaussian hypothesis...
详细信息
We consider the problem of selecting a subset of p out of n sensors for the purpose of event detection,in a wirelesssensor network（WSN）.Occurrence of the event of interest is modeled as a binary Gaussian hypothesis *** this case sensorselection consists of finding,among all（~n） combinations,the one maximizing the Kullback-Leibler（KL） distance between the induced p-dimensional distributions under the two *** exhaustive search is impractical if n and p are large,as the resulting optimization problem is *** propose a suboptimal approach with computational complexity of order■（np）.This consists of relaxing the 0/1 constraint on the entries of the selection matrices to let the optimization problem search over the set of Stiefel *** finding the Stiefel matrix is a nonconvex problem,we provide an algorithm that is guaranteed to produce a global optimum for p = 1,through a series of judicious problem *** case p > 1 is tackled by an incremental,greedy *** obtained Stiefel matrix is then used to determine the sensorselection matrix which best approximates its range *** simulations are used to assess near optimality of the proposed *** also show how the proposed approach performs better than exhaustive searches once an upper bound on the computation time is set.
暂无评论