Abstract: The invention is made of an apparatus which incorporates a sensor-based means for stabilizing random access networks. In the preferred embodiment, a grid of sensors is used to gather energy measurements for analysis. In a preferred embodiment a neural network has been trained to estimate the number of colliding users in a given slot. This information is used to set parameters in a backoff algorithm so as to stabilize the network and minimize the delay experienced by users. The invention has the ability to locate users geographically within the network coverage area. This information can be used in conjunction with a steerable beam or an array of antennas to develop geographically-determined Aloha subchannels, further increasing the capacity of the system.