A flow-based optimization model for throughput-oriented relay node placement in wireless sensor networks

This work addresses the relay node placement problem in wireless sensor networks. We consider a scenario in which all sensor nodes stream data towards sink nodes. Additional relay nodes can be placed with the aim of optimizing overall network performance in terms of data throughput. We formalize the problem with a linear, mixed integer mathematical programming model. We include a number of constraints and penalties to closely model the wireless environment. When relay nodes can be placed anywhere, we define their possible locations using a discrete grid. Model solutions specify both where to place the relays and the paths for routing. Through extensive simulation experiments, we compare model solutions against a state-of-the-art dynamic routing protocol to assess the quality of the routes, and against a relay node placement heuristic to evaluate relay positioning. To tackle the computational complexity, we also propose and study the effect of different strategies for determining the grid resolution. Additionally, an experimental validation carried out in a real testbed shows that the computed solutions clearly increase network performance by enabling the reception of larger number of data packets at the sinks and determining a fair QoS distribution among the nodes. Finally, we propose an on-line application, in which the model is built and solved on-demand, to adapt to changes in traffic patterns induced by external events, and the results are rapidly spread throughout the network and used to modify relay node positioning and routing paths.