Contribution to the optimal design and performance improvement of wireless mesh sensor networks

Resumen

In the last few years, Wireless Mesh Sensor Networks (WMSNs) have emerged as an enabling technology for providing potential pervasive computing services in next-generation wireless networks. Although WMSNs are not new conceptually, it was not until recently that the scientific, manufacturer and consumer communities have paid special attention to this technology. In stark contrast to traditional wired/wireless technologies such as Ethernet or Wi-Fi, WMSNs arise as a cost-efficient and reliable solution for long-term monitoring and control applications which allows the deployment of networks in diverse areas of interest regardless of its location (e.g., farming crops or forest areas of difficult access) and area to monitor (from a few square meters or hectares to several kilometers). WMSNs provide mainly these advantages thanks to its mesh capability, which guarantees robustness, scalability, reliability, security, interoperability, self-organization, energy-efficient design, and ease of maintenance, among others; through networks composed of hundreds or even thousands of small-sized, low-power, and low-cost wireless sensor devices. Therefore, considering the potential of this technology on a multiplicity of market segments, comprising automation and control (home, building and industrial), environmental surveillance, precision agriculture, and health services, among others, the present Thesis contributes in the field of WMSNs through new contributions and the performance enhancement. In this context, we first review the most relevant and up-to-date WMSNs standards and proposals that have emerged in the WMSN market in the last few years. To this end, we start by identifying the most significant requirements influencing the performance of WMSNs. The goal is to perform a deep study of each proposal in accordance with the different WMSN requirements in order to discern their main advantages and shortcomings. Then, the group of proposals is re-examined following diverse design guidelines in order to aid the end users and developers to select the approach which best adapts to the necessities of their application. Among all of the proposals reviewed, we highlight the IEEE 802.15.5 standard, a recent recommendation which fully satisfies the greatest number of WMSN requirements and design guidelines. In this regard, a description of IEEE 802.15.5 is provided. Concerning the IEEE 802.15.5 standard and, after a thorough study of the scientific literature, we realized that, to the best of our knowledge, most of the functionality provided by this specification is not evaluated. For this reason, since energy efficiency is one of the main design issues in WMSNs, we focus our interest on the energy-saving mechanisms denoted as Synchronous Energy Saving (SES) and Asynchronous Energy Saving (ASES) of the IEEE 802.15.5 standard. On the one hand, the SES mode is planned to support, thanks to its synchronous character, applications with strict temporal and bandwidth requirements (e.g., delay-sensitive applications). On the other hand, the ASES mode is conceived as an energy-saving mechanism to support the majority of traditional Wireless Sensor Network (WSN) applications, which employ an asynchronous time schedule for data transmission at very low traffic generation rates. Under these premises, we start by studying the performance of the SES mode. Simulation experiments reveal a deficient synchronization scheme, which provokes variable delays in the dissemination of information and reduces the network lifetime, thus limiting the full exploitation of the SES mode. To deal with this concern, we propose a new synchronization algorithm, denoted as HIgh-PErformance SYNchronization algorithm for wireless mesh sensor networks (HIPESYN), which is fully integrated into the IEEE 802.15.5 standard. HIPESYN is thoroughly evaluated by means of analysis and simulation, and its results carefully discussed. Results demonstrate that HIPESYN supports intensive bandwidth applications in a much better way than using the original synchronization algorithm. We continue our research with the study and assessment of the ASES mode in an environment where the so-called hidden and exposed terminal problems are considered. These phenomena are two of the current major problems which restrict the full exploitation of wireless communications, especially on the WMSN field because of the considerable number of energy-constrained devices forming the network and the multiple path alternatives connecting any arbitrary source-destination nodes. In this regard, results from simulation and test bed show that the hidden and exposed terminals produce a sharp degradation of the network performance and increase energy consumption, which can even jeopardize the ASES mode operation. For this reason, we present a comprehensible and deep review of the most relevant proposals that cope with the hidden and exposed terminal issues on WMSNs, dividing the set of proposals in four groups: time division multiple access (TDMA) protocols, multi-channel switching techniques, code division multiple access (CDMA) protocols, and directional antennas. Finally, a set of these approaches is evaluated in order to find out those techniques most suitable for their future integration into the ASES mode. Following with the concern about the hidden terminal phenomena on WMSNs, we propose an analytical model that mitigates this effect, simultaneously minimizing the sum of all the collision times in the network due to hidden terminals and maximizing the network lifetime. To this end, our analysis follows the rules of the multi-objective formulation and the goal programming technique, which model and compute the problem thanks to constraints that consider the most relevant premises referred to the ASES mode and hidden nodes. It is important to note that, to the best of our knowledge, these mathematical techniques have never been employed in the WMSN arena. Additionally, we also propose a MULti-channel TIme-scheduled algorithm for the HIdden Terminal problem avoidance (MULTI-HIT) which achieves a similar network performance in terms of throughput and network lifetime to the values returned by our analytical approach. Analytical and simulation results are presented and carefully discussed, showing the effectiveness of both proposals. Next, we focus on the Wireless Video-based Sensor Network (WVSN) field, a special type of WSN intended to provide added-value services (e.g., tele-surveillance) to the traditional WSN applications. Our goal is to study how to achieve the maximum network lifetime, and simultaneously satisfying the best aggregate throughput for any arbitrary network deployment (e.g., cluster, tree, grid, random), but having its major impact on mesh topologies in order to later, enable multimedia services in future WMSN deployments. To this end, we propose a planning model based also on multi-objective formulation and goal programming, which results in a more accurate solution than using the current optimization techniques applied both to WVSNs and traditional WSNs. Furthermore, we contribute with an OptimaL LOad BAlancing AlgorithM which ensures a fair traffic load distribution per link during the network operation and matches the numerical values obtained by our planning model. Simulation experiments and a real test bed deployed in a trial environment validate both our LOAM proposal and the planning model. Finally, two appendixes complete the present Thesis with some collateral contributions. Appendix A describes the implementation of the main functionality of the IEEE 802.15.5 standard for TinyOS, and offers users and developers some useful guidelines on how program their own applications using our software modules. In its turn, Appendix B provides implementation details of an application developed under TinyOS that enables video transmission using real wireless sensor devices.