Enhanced collision avoidance mechanisms for wireless sensor networks through high accuracy collision modeling

Abstract

Wireless channel and multi-hop communications cause a significant number of packet collisions in Wireless Sensor Networks (WSNs). Although a collision may cause packet loss and reduce network performance, low-power wireless transceivers allow packet reception in the presence of collisions if at least one signal can provide a sufficiently high power compared with other signals. Therefore, with respect to the large number of nodes used in WSNs, which necessitates the use of simulation for protocol development, collisions should be addressed at two layers: First, collisions should be modeled at the physical layer through a high-accuracy packet reception algorithm that decides about packet reception in the presence of collisions. Second, collision avoidance mechanisms should be employed at the Medium Access Control (MAC) layer to reduce packet losses caused by collisions. Unfortunately, the existing packet reception algorithms exhibit low accuracy and impede the development of efficient collision avoidance mechanisms. From the collision avoidance perspective, existing contention-based MAC protocols do not provide reliable packet broadcasting, thereby affecting the initialization performance of WSNs. In addition, despite the benefits of schedule-based MAC protocols during the data-gathering phase, the existing mechanisms rely on unrealistic assumptions. The first major contribution of this work is CApture Modeling Algorithm (CAMA), which enables collision modeling with high accuracy and efficiency at the physical layer. The higher accuracy of CAMA against existing approaches is validated through extensive comparisons with empirical experiments. The second major contribution includes mechanisms that improve the reliability of packet broadcasting. In particular, adaptive contention window adjustment mechanisms and the Geowindow algorithm are proposed for collision avoidance during the initialization phases. These mechanisms considerably improve the accuracy of the initialization phases, without violating duration and energy efficiency requirements. As the third major contribution, a distributed and concurrent link-scheduling algorithm (called DICSA) is proposed for collision avoidance during the data-gathering phase. DICSA provides faster slot assignment, higher spatial reuse and lower energy consumption, compared with existing algorithms. Furthermore, evaluating DICSA within a MAC protocol confirms its higher throughput, higher delivery ratio, and lower end-to-end delay.