Welcome - CANCCOM - Complex Adaptive Networks for Computing and Communication

Our goal is to develop performance analysis techniques for emerging network architectures. The goal is to provide information about performance, guidelines for design, and optimization recommendations for various SONs. Our plan is to suggest new models and algorithms and to characterizing their properties through applications of newly developed mathematical theorems. We plan to investigate the capacity and delay in wireless mesh networks. We will explore the impact of scalability. Many performance metrics in telecommunication networks are characterized through rare event probabilities, for example the loss probability. The characterization becomes very challenging for a typical SON since the dimension of the state space is usually very large. The large deviation technique is one of the popular methods to address this issue. However, its application to complex systems such as SONs is still too challenging. Based on newly developed results of the matrix analytic method, our goal is to propose easy-to-apply methods for characterizing stationary tail asymptotics in various SONs.

Queuing theory and applied stochastic processes are areas that have attracted great interest from the research community for the past several years. New network technologies appear having different operational principles and characteristics from traditional architectures, thus presenting problems that may not fall into the categories that were already investigated. We will concentrate on problems posed by new wireless network architectures. In particular we will focus in the area of a) fourth generation (4G) cellular systems and their interconnections to Internet backbones and b) sensor and ad-hoc networks based on IEEE 802.15/Zigbee protocols. New cellular 4G wireless networks are introducing a variety of new problems since they are expected to carry apart from voice other traffic types such as (semi-)real time video as well as data. The fact that the wireless channel has entirely different properties compared to the traditional wired channels has created a variety of problems that cannot be solved using the existing performance analysis techniques. Our focus in this area is to perform resource allocation and quality of service control using stochastic optimization (dynamic programming and Markov decision processes) and compare with heuristic/easy to implement controls. In the area of ad-hoc and sensor networks, we would like to not only perform theoretical research but also develop customized wireless/sensor network platforms. Chipsets and evaluation boards for such platforms are very affordable and can be easily introduced and used in an academic environment. Furthermore, graduate students will be greatly be benefited from practical hands-on experimentation. We envision sensor network platforms that will be independent of legacy protocols, such as TCP/IP, highly adapted to the needs and simple to deploy on a large scale. Sensor networks will be developed for monitoring goods, people and vehicles. Off-the-shelf components will be used and assembled together. Both the processing part and RF part will be addressed, as mix of hardware and software.