Improving the performance of wireless sensor networks using directional antennas
Javier Schandy
PhD thesis from Universidad de la República (Uruguay). Facultad de Ingeniería. IIE
Advisor(s): Thiemo Voigt, Leonardo Steinfeld
- Aug. 2020
Research group(s):  Microelectronica (gme)
Department(s):  Electrónica
Download the publication : Sch20.pdf [7.5MB]  

Abstract

Over the last decades, lots of new applications have emerged thanks to the availability of small devices capable of wireless communications that form Wireless Sensor Networks (WSNs). These devices allow sensing, processing, and communication of multiple physical variables while keeping a low power consumption. During the last years, most of the research efforts were spent on the development and optimization of wireless communication protocols, aiming to maximize the reliability of the network while achieving the lowest possible power consumption. In this thesis, we study how to improve the performance of these WSNs by using directional antennas. Directional antennas can provide a higher gain and reduce the interference with other nodes by concentrating the radiated power in a certain direction. We present the different kinds of directional antennas available for WSNs, and we select the 6-element SPIDA antenna as a case of study. We present an electromagnetic model of this antenna, and we incorporate it into the COOJA network simulator. We report the first complete characterization of this antenna, including the radiation pattern and S11 parameters. The characterization shows that the antenna has a maximum gain of 6.8 dBi, a Half-Power Beamwidth (HPBW) of 113° and a module of S11 parameter of -7.5 dB at the central frequency (fc = 2.4525 GHz). We also present a novel way to optimize the antenna without changing its design by isolating multiple director elements. We show that with this technique, the performance of the antenna can be improved in terms of maximum gain, narrower HPBW, and a lower module of the S11 parameter without making any changes in the antenna itself. We evaluate the impact of supporting directional communications in the different layers of the network stack. We analyze the different challenges that arise and propose optimizations to overcome them in order to take advantage of the benefits of directional communication. We present an analysis of the state-of-the-art in neighbor discovery protocols for WSNs with directional antennas, and we propose, implement end evaluate two novel fully directional protocols: Q-SAND and DANDi. We compare both of them with SAND, a fully directional neighbor discovery protocol. DANDi is a fully directional asynchronous and dynamic neighbor discovery protocol where the contention resolution relies on a collision detection mechanism. To the best of our knowledge, DANDi is the fastest neighbor discovery protocol for WSN with directional antennas, with the additional advantage of being able to discover every reliable communication link in a network without requiring any prior information of the network topology. We combine the directional neighbor discovery protocol with MAC and routing optimizations in order fully take advantage of the benefits of using directional antennas. We focus on convergecast, a typical data collection application where every node sends packets periodically to a sink node. We present DirMAC, a novel MAC protocol that fully supports directional communication, together with four different heuristics to optimize the performance of the protocols. One of these heuristics has the added major benefit of being completely distributed and with no need for offline processing. Our evaluation shows that optimizations at both the MAC and routing layers are needed in order to reap the benefits of using directional antennas for convergecast. Our results show that the performance of the network can be greatly improved in terms of packet delivery rate, energy consumption, and energy per received packet, and that we obtain the largest performance improvements in networks with dense traffic. Simulations with different node densities show that when using directional antennas the PDR increases up to 29%, while energy consumption and energy per received packet decreases by up to 55% and 46% respectively. Experiments with real nodes validate these results showing a significant performance increase when using directional antennas in our scenarios, with a reduction in the RDC and EPRP of 25% and 15% respectively, while maintaining a PDR of 100%.

Additional data

""

BibTex references

Descargar BibTex bibtex

Other publications in the database

» Javier Schandy