Multihop wireless networks can offer a wide variety of important applications that could have a substantial impact throughout society as mobile devices become more and more critical to our day-to-day lives. However, the infrastructure of such networks is vulnerable to many types of network dynamics such as node mobility, potential attacks, and misbehavior. This research aims to develop models and algorithms for a deeper understanding of multihop networks when multiple failures are present. Specifically, this project includes three thrusts under the theme of resilience-to-failure: (1) the design, modeling, and analysis of the impact of multiple failures based on the Semi-Markov Process; (2) the design, evaluation, and implementation of a new mechanism in order to minimize the impact of neighboring node failures; (3) the study of distributed optimization algorithms for hierarchy and cluster formation. The expected results, such as models of node behavior and failures, along with techniques for topology discovery and topology optimization, promise significant scientific impact by presenting a theoretical framework for the design and implementation of network algorithms. The societal impact of this research lies in the wide deployment of portable devices for which communication failures may affect human welfare. Resilience-to-failure, as one of the areas that integrate multi-disciplinary characteristics, is used to initiate innovative curriculum and to advance the integration of research and education. In addition to traditional publications, informative and instructional materials will be developed and posted on the web in a timely manner for fast dissemination of results and education in the developed techniques.
The goal of this research project is to develop a theoretical foundation for large-scale deployment and applications of multi-hop wireless networks, which is a new technology that enables wireless device to communicate directly among each other. In the past decade, multihop wireless networking has gone through a tremendous evolution because of its potential applications in our daily lives, such as phone-to-phone information exchange, vehicle-to-vehicle communications for intelligent transportation and accident notification, as well as tiny medical devices for healthy monitoring and reporting, and more. However, pushing such an advanced technology to commercial use has been hindered by several issues. For instance, routing protocols are critical to enabling a device-to-device communication, and many solutions have been developed for high throughput by reducing complexity of the protocols as well as by decreasing energy consumption for power saving. The challenge is that basic functions of routing protocols can easily be manipulated by individual users, either because of selfish intention or malicous attacks, which will dramatically undermine the deployment of this new technology. The challenge may become even more complicated when users with wireless device are mobile, moving from indoor to outdoor, and to vehiculars, along with a variety of mobile applications in our daily lives. For example, we may have questions like ``Can I use my smart phone to exchange music with my friends with no charge?'' and ``What security configurations I need to send emails, and make bank transactions?'' and more and more. Therefore, we resort to fundamental research issues to important, yet seemingly simple questions, the following issues: (1) We are motivated to develop a new mobiity model, which can be used to analyze properties and performance of wireless networks and understand how these characteristics will benefit mobile applications. This model, in particular, overcomes the limitations and drawbacks that have been found in earlier literature. (2) We focused on the modeling and analysis of the impact of node misbehaviors to network connectivity of slef-organized mobile networks, which has been rarely studied before. As a conclusion, besides mobility-induced failures, node misbehaviors can cause node isolation problem as well, which impacts the network connectivity significantly. Our work also provides a deeper understanding to network performance evaluation and multiple failure detection in the presence of node misbehaviors. (3) We studied the joint effects of radio channels and node mobility on link lifetime and its properties. We have found that radio channel characteristics may predominate the link performance for slower mobile nodes, while node mobility dominates the link performance for faster mobile nodes. The link lifetime, which demonstrates the quality of communication links between mobile devices, can be effectively approximated by exponential distribution. As a fundamental study, our analytical results and simulation findings on link dynamics can be readily applied to system design such as topology control and routing optimization. (4) We proposed and implement a Self-TunEd Performance and Protection (STEP2) manage- ment system for wireless LANs, in which an adaptive and dynamic scheme is used to switch network protection (security policies) for optimizing network performance. It is found that STEP2 is very affective in achieving the balance for the systems that require a tradeoff between performance and protection. For example, we observe that STEP2 incurs upto 40% less packet losses and 30% less per packet delay in different scenarios. (5) We tackled the distorted forwarding problem by achieving an optimal resilient topology over the given wireless network. Based on our analytical results of resilient capacity of multihop networks, we designed a new distributed topology control protocol, PROACtive, to enable every node to build up its own cooperative neighborhood dynamically. (6) We asked a generic networking question: if a node can fail, say with a certain probability or distribution, because of any reasons, from electronic fault to communication interruption, how long will the entire network be disconnected? An interesting result is that the network with heavy-tailed survival functions (e.g., Pareto) is no more resilient to random failures than the network with light-tailed ones if the expected node lifetimes are identical. Finally, we find that the network is highly resilient to random failures in terms of a ``graceful'' devolution of the giant component, which includes most of the nodes in a network. (7) We have found that wireless mesh networks can provide satisfactory communication support in the smart grid including the most time critical commu- nications. The communication, scheduling and routing schemes investigated in this paper also provide example solutions for the wireless mesh network design and operation as well as the delay performance evaluation references for more advanced studies in the future. (8) Our proposed framework, Greenbench, which contributes a new cross-domain simulation platform that includes an underlying power system overlayed by a communication system. In such a way, we are able to capture the impact of cyber attacks in power systems in real-time, unlike networking simulations or physical systems. We aim to study the consequences of data-centric attacks rather than manipulation of communication protocols. Our study is able to demonstrate the direct impact of security attacks at the power system level, either due to compromised smart meters in Automatic Metering Infrastructure or DoS attacks to messages in transmission.
