Complex Self-Organizing Systems

Even though I have a deep interest in this field, this work is a relatively new part of my research portfolio. There are several organisms in nature that self-organize to exhibit sophisticated emergent behavior that stems from simple interactions of each of its components. There are several artificial systems that exhibit similar behavior. For instance, computer networks are complex systems whose components make independent decisions through interactions with their local neighbors. Even though the behavior of each individual network component is simple, the network itself exhibits sophisticated behavior. Such systems have been examined using the theory of self-organization. We are attempting to use this approach to study the behavior of several artificial systems, such as, nano-communication, sensor networks, computer networks, and engineering optimization.

Mutual Fund Complexity

This work entails estimating the impact of complexity on the performance of mutual fund portfolio performance. In this research, complexity of decision-making is examined in the context of portfolio management. Since increased complexity entails additional costs, it is important to ensure that the cost of added complexity is reflected in improved performance. This research examines the hypothesis that the additional sophistication of analysis in portfolio management leads to improved performance. We use historical data on portfolio performance and relate it empirically to the complexity of the decision-making styles. The results demonstrate a strong negative correlation between portfolio performance and the level of complexity.

Nano-Sensor Networks (Not to Publish)

Technological advances in nanosciences are allowing creation of miniature inexpensive sensors that hold promise in several areas including: protecting the environment, reducing engine emissions, delivering measured doses of drugs to the body, and providing advanced sensing capabilities in modern battlefields. These advances have been very slow to realize mainly because of the problem of coordination among multiple sensors. In an ideal world, each sensor would have a power source, a processor, and its own transmission capability. In reality, however, to achieve miniaturization, processing and transmission capacity become seriously compromised making coordination from an external agent difficult. Each sensor is fairly weak, however collectively the sensors can perform complex tasks. To effectively use these sensors they need to self-organize and communicate with each other without external intervention. In conjunction with Dr. Michael Carpenter from the College of Nanosciences and Engineering and Dr. Stephen Bush, we have developed a novel idea for communication among nano-sensors using the theory of self-organization. The proposed work entails creating sensor networks that are endowed with self-organization capability such that simple sensors can collectively perform complex tasks. We will investigate the self organizing capabilities of multi-transduction sensor platforms which incorporate optical, electrical, acoustical, or combinations thereof.

Self-Organizing Transportation (Not to Publish)

In this research, we will develop a transportation system in which traffic lights communicate with each other and adapt their behavior in real-time to improve continuity of traffic and to minimize wait time. Traffic signals, though essential, can be often quite frustrating especially when you have to stop at an intersection and wait for a signal to activate even when there is no other car within miles at a crossroad. It can also be quite inefficient when the signal activates for a car while tens of cars are traveling at a high speed on the crossroad and are just seconds away from the signal. By communicating with each other, the lights can synchronize to minimize such disruptions and maximize the traffic flow with minimal wait time. This work would decrease the average wait time, reduce congestion and lessen green house emissions by reducing the idling time of vehicles. Due to its adaptive behavior this system would also be more robust and handle anomalies more efficiently. Such self-organized behavior observed in nature such as in social organisms like ants and bees as well as in physical phenomenon such as crystallization of liquids. Algorithms will be developed for self-organization traffic signals. These algorithms would not entail communications across all the signals in the system but just the signals in the immediate neighborhood. A simulation will also be developed to test the behavior of such a system as it is scaled and when anomalies happen like accidents, and road blockages.

Related Publications

  1. Bush S. F., & Goel, S., (Sept. 14-16, 2006). Graph Spectra of Carbon Nanotube Networks. Proceedings of the First International Conference on Nano-Networks. Lausanne, Switzerland.
  2. Goel, S., Brown, C. D., & Shawky, H. (2007). Complexity and the Performance of Investment Portfolios. Accepted February 2007 in the Advances in Investment Analysis and Portfolio Management.