Reliable and energy efficient cooperative detection in wireless sensor networks

The efficiency of a wireless sensor network (WSN) is dependent on the level of event detection reliability it achieves, and the associated consumed amount of energy. In this work, the potential performance improvement gained by using a routing metric that reflects the quality of links in terms of detection reliability is investigated. As a first main contribution, we derive the expression of such detection reliability-aware link metric considering a realistic multi-dimensional Gaussian autoregressive (AR) model to describe information correlation within the supervised field. Such routing metric quantifies indeed the participation of each link on a path in reducing the probability of error when making final decision at the fusion center (i.e., the sink node). As a second main contribution, we propose that only the new information among the sensed information by the sensor nodes be propagated to the fusion center instead of the whole raw data. As such, the number of packets traversing the network is considerably reduced, which leads to significant energy conservation. To achieve this in a relatively easy way, we use fast filtering to aggregate data information at intermediate nodes along the path to the fusion center. Finally and as a third main contribution, we balance detection reliability and energy-efficiency by including a weighted value of the energy consumption in the expression of the detection reliability-aware link metric. This cooperative routing approach is shown to yield significant benefits to the network, by either increasing the provided QoS (i.e., the offered level of detection reliability) or allowing the network to decrease its energy consumption compared to classical noncooperative routing schemes and the cooperative routing scheme with one-dimensional correlated field.