Radio Geometry-Aware Resource Allocation in Urban-Deployed Cell-Free LoRa Wireless Networks

Low-power wide-area networks (LPWANs) are ideally suited to provide flexible long-range (LoRa) wireless access networks for internet-of-things (IoT) devices in smart city applications. Considering the shadow fading of the large and closely located buildings in the urban environment, one of the key challenges in the LoRa network deployment is spreading factor assignment and power allocation. In this paper, a new radio geometry (RG)-aware resource allocation framework for cell-free LoRa wireless networks in urban areas is proposed. First, a stochastic optimization problem for the spread factor assignment and power allocation in a multi-gateway cell-free LoRa wireless network is formulated. Next, the radio geometry of LoRa wireless networks as a set of vectors, including the channel gain between each LoRa EN and multiple LoRa gateways is defined. Then, an unsupervised learning-based solution is used to partition the radio geometry vectors and find a solution for spread factor assignment. Next, the greedy algorithm is applied for the LoRa EN power allocation to maximize the network throughput. The practical measurement is done to model the line-of-sight (LoS) and non-LoS LoRa wireless channels for different real measurement scenarios over the campus of the Shahid Chamran University of Ahvaz. Then, our proposed resource allocation algorithm for the set of radio geometry vectors is evaluated in a given area of the campus of Shahid Chamran University. Our simulation results show that the collision ratio under our proposed RG-aware resource allocation algorithm is on average 20% and 5% less than the collision ratio under distance-based and SNR-based algorithms, respectively.