PARAMETER ESTIMATION FOR HIERARCHICAL CHANNEL PROFILING

This paper is a first step towards developing a channel model that will allow channel prediction in time and space. The wireless channel is described in time and space using a hierarchical model. At each point in time, a set of networked nodes measure their received signal strength (RSS), which is correlated in time and space according to a known statistical model. The channel state, a vector of channel characteristics, such as the path-loss exponent and shadowing power, is time varying and hidden from the network nodes. Based on this model structure, we propose an algorithm, which divides the recorded RSS into time segments, over which the channel state is constant, and also provides state estimates corresponding to each segment. The proposed model is subsequently fitted to indoor Wi-Fi measurements, obtaining meaningful segments, and providing insight on the time evolution of the channel parameters.