Existing Biologically Inspired Algorithm (BIA) approaches to Self-Organized (SO) networks lack a theoretical foundation and analysis. This work develops a framework to describe the types of problems that BIA can address in network control. First, this work establishes a theory of BIA network control. Here the PIs: a) broadly review the literature for BIAs that have been applied to solving control theory problems and develop a categorization of the methods; b) categorize the common problems in network control and the complexity of these problems; c) analytically demonstrate the potential of these BIAs in solving various network control problems; and d) formally describe BIA for control of wireless networks. Second, this work analyzes mutual interaction among layers in a BIA-based network. The PIs carry out an analysis of mutual interaction among layers in BIA-based networks. This includes a) analysis of interaction among layered SO-based control and b) establishment of a fundamental theory of layered architecture. Third, this work analyzes BIA-based control in wireless networks. Here the PIs conduct analyses of BIA-based control on wireless networks, to a) determine implementation issues of BIA-based control on real systems and b) perform a practical evaluation and analysis of BIA-based control through simulation and testbed implementation. This work is significant in that it provides a framing of BIA and provides a means for others to assess new approaches beyond wireless applications. The results of this work will be made available in conference and journal publications and through web sources.
This work examined the design and performance of self-organized network architectures. In self-organizing network architectures, the globally integrated behavior of the system emerges through mutual interaction among networks and nodes operating on local information. Biologically Inspired Algorithms (BIAs) are processes that mimic how nature solves problems and these BIAs often operate in a self-organizing manner. This research looked to answer such questions as: What kind of self-organization can be applied to which network control problems? How well do these algorithms and problems conform to the notion of distributed or local control? How well can we expect these algorithms to operate - both in terms of solving the given problem and in terms of computational overhead? How do these algorithms perform when implemented in actual wireless networks? The fundamental goal of this work is to provide a rigorous structure for assessing the general applicability of self-organized mechanisms for wireless network control problems. The work found that self-organized control can be applied successfully to a host of wireless networking problems and that the basis of many BIAs devolve to a simple set of control mechanisms [1]. These initial findings led the group to explore self-organization for various Mobile Adhoc NETworks (MANETs). An important result here was that self-organized control could be augmented with context information (e.g., movement and location) to improve throughput performance of MANETs and hybrid MANETs while reducing computational load, network overhead and power consumption [2]. Additionally, this research examined centralized control for hybrid MANET as a comparison to the biologically inspired self-organization control [3], and the project had a related publication examining distributing traffic information among self-organized groups of smartphones [4]. Lastly, the team recently wrote a number of pending publications, including some with our collaborators in Japan. The key contribution in these more recent publications is a mathematical basis for evaluating more complex control environments, such as those that include context information. An important part of this grant was fostering joint Japan/U.S. collaboration. The US and Japan team meet numerous times over the lifetime of the grant, including a 4 months stay by a Japanese graduate student in PI Sicker’s lab, and through this interaction the students completed an implementation and evaluation of a hybrid MANET on Android devices. In terms of an educational component, this grant supported Chenyu Zheng, who will defend his Ph.D. work this summer. His work explores formal analysis of self-organized MANETs and includes an experimental component where these protocols are tested on actual mobile devices. The findings of this research have broader impact in a number of areas. First, the survey work highlights areas worthy of additional research. Second, this research shows specifically how self-organized control could aid in improving critical mobile communications for such purposes as public safety, search and rescue, and national defense communications. Third, this work successfully brought together researchers from different countries to collaborate. Publications: [1] Chenyu Zheng, Douglas Sicker. "A Survey on Biologically Inspired Algorithms for Computer Networking, IEEE Communications Surveys and Tutorials Journal (IEEE COMST), 2013. [2] Chenyu Zheng, Douglas Sicker, Lijun Chen. "Self-Organized Context-Aware Hybrid MANETs, 10th International IEEE Annual Conference on Wireless On-Demand Network Systems and Services (WONS 13), 2013. [3] Murad Kaplan, Chenyu Zheng, Eric Keller. "WASP: A Centrally Managed Communication Layer for Smart Phone Networks", 10th USENIX Symposium on Networked Systems Design and Implementation (NSDI ’13), 2013. [4] Jun Du, Chenyu Zheng, Zhang Zeqi, Zhai Zhongqiang, Yu Yang, He Nengqiang, Douglas Sicker, Yong Ren. A Smartphone-based Traffic Information Service Platform for Pedestrian and Bicycle Systems, 15th International IEEE Annual Conference on Intelligent Transportation Systems, September 2012.
