The objective of this research is to offer a fundamental study of video communications over dynamic ad hoc networks. The approach is to exploit recent advances in multiple description coding, multi-path routing, packet scheduling, and per-node power control across the layers of the protocol stack. Specifically, there are three main thrusts in the research activities: (1) cross-layer modeling and theoretical study; (2) design of distributed algorithms and protocols; and (3) testbed prototyping and validation.
The intellectual merit of this project includes the following components. First, there is limited research pursuing a holistic consideration of video transport under the impact of multiple lower layers. Such efforts, however, are necessary to optimize video performance. This research will thus offer new cross-layer video optimization models and theoretical results. Second, the efforts on the development of distributed algorithms and protocols will offer not only practical operational algorithms, but algorithms that can approach the theoretical limits. Finally, the efforts on testbed prototyping will offer deeper insights into the challenges associated with video communications over ad hoc networks as well as valuable feedback to the theoretical and algorithmic research.
The broader impacts of this project include: (1) the development of new educational materials to bridge the gap between software applications, networking, and wireless communication, and (2) research experience for undergraduates and special opportunities for underrepresented minority students.
The goal of this project was to address some key barriers that hinder video communications over dynamic ad hoc networks. Prior to this project, there was a lack of research efforts on addressing how to sustain video quality in ad hoc networks under extreme environments (e.g., mobility, node failure, jamming and limited battery energy). To address the video quality issue, we exploit a new video coding technique called multiple description (MD) coding and multipath routing. MD coding is capable of encoding a video into multiple independent streams (or descriptions) such that any subset of these streams can be combined at the receiver to reconstruct the original video. In the worst case, information received from any one video stream can be used to reconstruct and render a minimum acceptable video quality. The quality of the reconstructed video improves as more streams are received and combined. The characteristics of MD coding, and the path diversity of an ad hoc network, renders a unique approach of combining MD coding and multipath routing to optimize video quality. To address the energy constraint at each wireless node, we exploit a new technology called wireless energy transfer (WET) based on magnetic resonant coupling. By employing a mobile vehicle to charge each wireless node with the WET technology, we found that the energy bottleneck at each node in the network can be effectively removed, allowing energy-intensive video communications to be supported in such a network. We have developed new mathematical models and theoretical results for video communications over ad hoc networks by exploiting advances in video coding and energy transfer. By unifying algorithms at multiple layers across the protocol stack, we were able to develop joint optimization models for MD video communications in a dynamic ad hoc network. Based on such optimization models, we developed theoretical results and validated the advantage of a unified cross-layer approach over other existing approaches. For the energy problem, we studied the scenario where a mobile vehicle is employed to carry a wireless charging station and periodically visit and charge the wireless nodes in the network wirelessly. We offered necessary and sufficient conditions to ensure energy at each node can be replenished in a timely manner so that energy-intensive video communications can be effectively supported in such networks. Guided by the new mathematical models and theoretical results, we developed distributed algorithms and protocols for MD video over dynamic ad hoc networks and for wireless energy transfer. We conducted extensive simulation studies and showed that our distributed algorithm can approach the desired network theoretical limits. In addition to developing theoretical foundation, we also developed a prototype on a testbed for validation. We implemented our cross-layer video routing algorithms by leveraging off-the-shelf ad hoc routing algorithms. Our testbed was able to emulate network topology, link success probability, and link bandwidth. Using the testbed, we conducted a series of experiments and concluded that our proposed approaches offer a viable solution to address video communications in a dynamic ad hoc network. An integral component of this project is graduate student education. One graduate student was supported under this project. Over the project years, the student was trained in courses on Optimization, Simulation, and Wireless Networks to meet the technical challenges of the project. The student was also trained by the investigators regarding research methodology, writing, and presentation skills. As a result, the student was able to make important contributions to this project. Another female student also benefited from participating in the project. She was mentored by the PI and completed a M.S. degree in Computer Engineering. She has since decided to pursue a Ph.D. degree on wireless networks at the PIs’ institution. The PIs have co-authored a new graduate textbook titled "Applied Optimization Methods for Wireless Networks," which is published by Cambridge University Press (Spring 2014). This is the first textbook that teaches how various optimization methods can be applied to solve challenging wireless networking problems. Two chapters in the book were drawn from the research problems in this project. This book offers students a unique perspective on how optimization methods can be designed to solve practical wireless networking problems. The PIs have been disseminating research findings through major conferences and journals. They gave a number of research seminars at universities worldwide that described research findings from this project. The PIs have been hosting many visitors from academia, government, and industry to their institution and have been offering briefing of their research activities. The activities and findings from this project were an important component in these information exchange briefings.
