Collaboration among computational nodes, whether they are wireless or connected, mobile or stationary, is almost always beneficial. Practicalities do, however, limit the level of collaboration possible. In this paper, the authors consider wireless sensor networks (WSNs), aiming to extend remote sensor node battery life. They propose doing this by reducing the energy expended in data transmission through compression of the payload using data in prior transmissions by colocated sensors.
The authors start by showing that viewing cluster-based WSNs as communicating using point-to-point channels unnecessarily limits the scope of any analysis, as nodes tend to have omnidirectional antennas. The motivation offered for compression is that spatially colocated sensors tend to exhibit high degrees of data correlation.
The authors factor in both the energy consumed for reception of peer transmissions and the energy expended in compression of the data. By formulating an energy-consumption function based on the location of sensor and relay nodes, the authors maximize the time before the next sensor runs out of power. Two optimization schemes are discussed in detail: the first is based on linear programming; the second is a heuristic algorithm that, despite being significantly simpler, exhibits very good results under simulation testing.
The simulation analysis is thorough considering such aspects as: spacial sensor layout; compression based on more than one peer’s transmission; time to nth sensor’s death; compression ratios; cluster sizes; and energy consumption rates for data reception. Even packet header overhead and data loss are analyzed.
Despite the supposed potential for improvement, at only the cost of increased complexity, the authors fail to consider some of the practical implications. Is it not cheaper to double the battery capacity by removing the cost of reception capability at the remote sensors? This is pertinent, as neither the analysis nor examples necessitate bidirectional communication from the remote sensors.