A Distributed Reception Architecture for Low-Power Wide-Area Networking

A large number of pervasive sensors will manage and optimize the infrastructure in future smart cities. Ideally, we will be able to sense everything from bridges, buildings and firetrucks, all the way to trash cans, bicycles and street lights. The growing field of low-power wide-area networking (LPWAN) looks at the challenge of wirelessly communicating with a large number of low-power, simple devices over long periods of time. These city-wide networks need to provide two important functions: extracting data from sensors and keeping track of where they are located. We explore the capabilities of these networks by deploying and evaluating our own OpenChirp LPWAN network around the Carnegie Mellon University campus. Though promising, these networks still suffer from problems of coverage, conservation of device battery and the lack of an ability to accurately localize devices. This thesis will explore the design space and potential of using tightly coordinated distributed gateway receivers to efficiently decode data, as well as to locate the transmitting devices. It is challenging to develop distributed radio systems that utilize physical layer signals without overwhelming the backhaul network. We first introduce a coherent combining system, called Charm, that improves network coverage, data rates and battery life of deployed devices by selectively collating receptions from multiple receiving gateways. In an indoor environment with high multipath, accurate time synchronization at nanosecond accuracies is challenging, but plays an important role in enabling both localization and coherent combining. Thus, we develop a time synchronization platform, called Pulsar, for nanosecond-scale synchronization of radios. Next, we show how we can extend our distributed reception system with time-synchronized receivers to improve device localization through the use of time-difference-of-arrival features in challenging urban environments. With multiple receivers that are accurately time synchronized, we see the potential to significantly improve LPWAN performance and effectively localize transmitting devices