Accurate Corruption Estimation in ZigBee under Cross-Technology Interference

Abstract

Cross-Technology Interference affects the operation of low-power ZigBee networks, especially under severe WiFi interference. Accurate corruption estimation is very important to improve the resilience of ZigBee transmissions. However, there are many limitations in existing approaches such as low accuracy, high overhead, and requirement of hardware modification. In this paper, we propose an accurate corruption estimation approach, AccuEst, which utilizes per-byte SINR (Signal-to-Interference-and-Noise Ratio) to detect corruption. We combine the use of pilot symbols with per-byte SINR to improve corruption detection accuracy, especially in highly noisy environments (i.e., noise and interference are at the same level). We extract pilot symbols by leveraging protocol signatures. In addition, we design an adaptive pilot instrumentation scheme to strike a good balance between accuracy and overhead. We implement AccuEst on the TinyOS 2.1.1/TelosB platform and evaluate its performance through extensive experiments. Results show that AccuEst improves corruption detection accuracy by 79.4 percent on average compared with state-of-the-art approach (i.e., CARE) in highly noisy environments. In addition, AccuEst reduces pilot overhead by 83.7 percent on average compared to the traditional pilot-based approach. We implement AccuEst in a coding-based transmission protocol, and results show that with AccuEst, the packet delivery ratio is improved by 22.1 percent on average.