Energy efficiency and scalability continue to be key considerations for the development of low-cost wireless networks for meeting the needs of the emerging world of the Internet of Things (IoT). Recent developments in low-power wide area networks (LPWAN) promise to meet these requirements by achieving long communication ranges at low data rates without increasing the energy cost.
Consequently, LPWAN technologies are rapidly gaining prominence in the development of IoT networks in comparison to legacy WLANs. LPWANs address the challenges of legacy wireless technologies that use multi-hop mesh networking for increasing connectivity and coverage. Long Range (LoRa) technology is receiving increasing attention in recent years for addressing the challenges of providing wireless connections to a large number of end devices in the field of IoT. LoRa has become the most prominent LPWAN standard due to its long transmission range, low power consumption, and large network capacity. Despite these benefits, LoRa networks may not be able to achieve their full potential unless additional improvements are achieved in the network scalability domain. Specifically, the probability of success under heavy network traffic loads or a large number of end devices needs to be improved.
In this dissertation, we present the causes of performance degradation of LoRa networks and propose several approaches to enhance their performance. Next, we present a novel framework to employ AI tools to make IoT applications smarter. The effectiveness of all the proposed approaches is validated using mathematical analysis as well as via simulations thereby creating the basis for further research in this area.