Design and Performance Analysis of MEC-Aided LoRa Networks With Power Control

Abstract

In this paper, we propose a single-cell mobile edge computing (MEC) assisted long-range (LoRa) network with power control, which includes a gateway, a MEC server, and many randomly distributed end-devices (EDs). In the proposed system, task packet messages can be computed locally through EDs and offloaded to the MEC server for edge computation; both computations are performed in parallel. Following the stochastic geometry theory, we adopt the homogeneous Poisson point process (PPP) to capture the randomness of the EDs' position and model the interference devices as PPP under the pure ALOHA. In this model, we consider both the interference caused by the same spreading factors (co-SF) and that caused by different spreading factors (inter-SF) during task offloading. Furthermore, we derive precise and approximated expressions of the computation offloading success probabilities for the proposed network, which are then verified by simulations. This is followed by the analysis of the impact of the power control on the network performance of the proposed network. The results reveal that power control can considerably improve the performance in the low-density EDs scenario, as well as slightly improve in the high-density EDs scenario. Finally, we investigate the performance of the proposed network by comparing three SF allocation schemes-namely, exponential windowing (EW), equal-interval-based (EIB), and equal-area-based (EAB) schemes. The results reveal that we can improve the performance by assigning a lower SF for a large number of EDs.