Wireless networks enable people to enjoy mobility and connectivity at the same time. An all-wireless mesh network can provide city-wide coverage, with key access points being connected to the Internet. The low cost of these networks makes them attractive for city police and emergency services, community networking, and basic connectivity in developing countries.
Mesh networks introduce fundamental questions of how shared resources are divided among competing users. Cooperation between nodes of the network is necessary for users to have global access, yet bandwidth limitations induce competition. This competitive environment is enticing to malicious users, who can then strategically induce network conditions to achieve various objectives. Modeling and mapping wireless networks poses fundamental challenges that are essential for managing competition and protecting against intruders.
We have studied the wireless network modeling problem from the perspective of optimal rate control for the users of a wireless mesh network. We have developed a model that incorporates partial interference that yields a convex optimization problem when formulated using link rates, with numerical results illustrating significantly higher utilities as compared to models that treat interference as contention. We have also developed a first-principles model of wireless networks based on representing the perceived times during which a signal is sent as a random set. We have shown that the first-principles model reduces to the classical models under certain limiting conditions. Because the first-principles model incorporates a more complete model of the wireless network, we are able to use it as a benchmark to evaluate the performance of other rate controllers that are more practical to deploy.
Our current work explores the problem of mapping a wireless network and detecting interfering devices. We are demonstrating how it is possible to use the signal structure of the network, which is the open-loop causal relationships among signals, to represent causal relationships between the links in the wireless network. This builds directly on work that developed dynamical structure functions to represent biological systems whose internal states cannot be completely observed. We are developing a protocol that can construct the signal structure of a wireless network using perturbation experiments. The resulting map can then be used to detect intrusive devices that temporarily disrupt communication and impair performance.
This work is done in collaboration with the BYU IDeA Labs.
This work is supported by the AFRL under Grants FA8750-09-2-0219 and FA8750-11-1-0236.