Faster and better the promise of dynamic spectrum access: Invited talk

Abstract

As wireless networks continue to proliferate, the problem of limited bandwidth has become increasingly apparent. In response to this issue, there has been much significant and successful research addressing the question of how to use the available bandwidth more efficiently, and how to make wireless protocols more scalable. For those researchers focused on the foundations of wireless (or mobile) computing, this direction has, for example, encompassed research on contention resolution, data dissemination, and routing in dense networks; it has included research on finding low-degree overlay structures and schemes for data aggregation. For those researchers focused more on systems research, it has led to new MAC-layer protocols and even new radio designs. Overall, there has been significant progress, particularly in the case of fixed networks organized around static base stations. Dynamic Spectrum Access. Recently, a new idea has taken hold: dynamic spectrum access. The basic observation is that in a dynamic network, such as a mobile ad hoc network, it is hard to assign spectrum in a static manner that will remain efficient even as a network changes. An alternative to performing a centralized off-line computation that partitions the spectrum among wireless devices is for each device to dynamically determine which portion of the spectrum to use. The hope is that such a dynamic spectrum allocation will be more efficient than centralized static allocations. For example, when one portion of the spectrum becomes too crowded, devices will automatically move to less crowded portions of the spectrum. When new applications are deployed, existing applications adapt seamlessly to the increased spectrum usage. Challenges. There are many challenges in deploying the idea of dynamic spectrum access. First, there must be appropriate hardware: radios that can quickly move between different portions of the spectrum. Recent improvements to software defined radios (see, e.g., GNURadio) have made this increasingly feasible. Second, there must be a basic software platform that enables wireless applications to take advantage of these new radios, accessing the spectrum both to transmit and receive information, as well as to sense occupied and unoccupied regions. Again, recent research has made significant progress in this area. Finally, there must be algorithms that will take advantage of the available spectrum to perform computation faster, more efficiently, and more robustly. Algorithms for Mobile Networks. In this talk at the Workshop on Foundations of Mobile Computing, I will address this last area: algorithms for mobile networks equipped to access the spectrum dynamically. From an algorithmic perspective, there are at least three ways in which dynamic spectrum access may lead to better protocols. First, by reducing contention and multiplexing a computation across a broader portion of the spectrum, it may be possible to perform some tasks faster than would otherwise have been possible. Second, by using the spectrum more efficiently, it may be possible to reduce the energy usage of a protocol, making it more efficient. Third, by adaptively using more reliable portions of the spectrum, and avoiding lossy portions of the spectrum, it may be possible to develop protocols that are more robust to interference and noise. In fact, using variants of frequency hopping techniques, it may even allow applications to tolerate malicious attackers that are actively attempting to disrupt an application. The main goal of my talk, then, is to discuss the question: how can we leverage the rapidly evolving technologies for dynamic spectrum access to make mobile wireless applications faster and better? To this end, I will review some of the existing work that may help us to address this question, focusing on research in multichannel networks. In closing, I will describe some of the problems we may face in making wireless networks more reliable. © 2011 Author.