Poster Abstract: The Utility of Wall-Blockage Modeling for Link Quality Prediction in Indoor IoT Deployments
Anitha Varghese, Shivam Vinayak, Anurag Kumar, Rajesh Sundaresan · 2020
We consider the problem of deployment of indoor multihop wireless networks for connecting sensors to a data collection station, in the context of Internet of Things (IoT) applications. The locations of the source nodes and the sink are fixed, and additional router nodes might be needed to create a connected network that provides the required quality of service (QoS). Practical constraints often dictate that these cannot be placed just anywhere, and so we assume that several potential relay locations are provided. The problem is then to design a multihop network connecting the sensors to the sink, using a minimal set of the potential relay locations, that meets the network QoS. A priori, the qualities of links terminating on potential locations are not known. We are interested in a predict-place-iterate approach for relay deployment. Thus, the quality of the deployed network depends on the quality of the link prediction model. In this work, we study the improvement in network deployment that is provided by including the number of intervening walls on the link, in addition to using link length, in the link prediction model. Our comprehensive study involving analysis, simulations and experimental validation demonstrates that including the number of walls in the link prediction model can lead to a larger probability of successful design, fewer router nodes, and fewer iterations until successful design.