Research On Low-Power Wireless Link Correlation

Recent advances in low-power wireless technologies have enabled the increased applications of low power wireless networks. In low power wireless networks, many recent works have demonstrated that packet receptions on adjacent links are correlated, which is different from the long held assumption that wireless links are independent. The correlation between packet receptions on different links originating from the same source node is termed as "link correlation". Link correlation has shown significant impact on wireless protocol performance, such as flooding, opportunistic routing and etc. Existing research works mainly focus on specific applications of link correlation, leaving unsolved the more general problems of link correlation measurement and link correlation-aware performance modeling. As a result, existing works suffer from inaccurate link correlation measurement and inaccurate performance models. To fill the gap, we focus on the three key problems on link correlation:measurement, link correlation aware performance modeling and link correlation aware protocol applications.(1) Link correlation measurement. Accurate link correlation measurement is the base for link correlation aware protocol optimizations. Existing works measure link correlation based on the receptions of beacons and lacks quantative model for link correlation, resulting in inaccurate measurement. To solve this problem, we first analyze the impacting factors of link correlation via tremendous experimnets. Then we build a characterization model of link correlation (CorModel) leveraging all impacting factors from both network layer and PHY layer. Based on CorModel, we propose a novel accurate and lightweight link correlation measurement scheme. The proposed scheme achieves much higher accuracy than existing measurement works. This work has been published in IEEE INFOCOM 2015.(2) Link correlation aware network performance modeling. Performance modeling is a fundamental building block for various applications, such as routing and simulation. Exising performance models fail to consider link correlation, resulting in inaccurate estimation of network performance. To solve this problem, we integrate link correlation as well as link quality and link temporal distribution into an accurate performance model (3DW model). Given certain wireless conditions,3DW model is able to accurately estimate the protocol performance. The proposed performance model provides general opportunities for both broadcast and anycast based network protocol optimizations. This work has been published in IEEE SECON 2014 and Sensors.(3) Link correlation aware data dissemination. We explore the applications of link correlation in the context of data dissemination protocol. Existing works fail to consider link correlation, and thus incurs inefficient dissemination. To solve the problem, we first analyze how link correlation affects the performance of dissemination process components:sender selection and transmission/coding strategies. Then we propose an accurate sender selection algorithm and an efficient coding strategy by exploiting link correlation information. Compared with existing dissemination works, the proposed dissemination protocol improves the performance in terms of completion time, energy consumption and transmission count. This work has been published in IEEE Transactions on Industrial Eletronics.