Skip to Main Content
In this paper, we describe an adaptive network/routing algorithm that facilitates both coherent and non-coherent event-based cooperative signal processing. The core of this algorithm is a distributed election procedure that produces one or multiple winners based on a context-dependent election metric. In scenarios where non-coherent signal processing techniques are applied, a central processing node is selected, and highly compressed sensor data is gathered for processing. Energy efficiency is improved by reducing algorithm overhead because the actual sensor traffic volume is light compared to the messaging overhead of the algorithm. For coherent processing, raw data streams must be relayed from each sensor to the central processing node, producing large data streams. A multi-winner election process is initiated first to select only a limited number of sensors that will provide the raw data; then a second election process uses an energy-based metric to find the optimal central processing node, whose location minimizes the total relaying cost. Simulation results are provided to demonstrate the inherent overhead-delay trade-off and compare the scalability of the algorithm under different scenarios.