Self-Evolvable Protocol Design Using Genetic Algorithms

Self-modifying protocols (SMP) are protocols that can be modified at run time by the computers using them. Such protocols can be modified at run time so that they can adapt to the changing communicating environment and user requirements on the fly. Evolvable protocols are SMP designed using Genetic Algorithms (GA). The purpose of this article is to apply Genetic Algorithms (GA) to design an evolvable protocol in order to equip communication peers with more autonomy and intelligence. The next-generation Internet will benefit from the concept of evolvable protocols. In this article, we design a Self Evolvable Transaction Protocol (SETP) with a GA executor embedded. We then use the Network Simulator (NS2) to evaluate this evolvable protocol module to demonstrate the feasibility of our new design approach. DOI: 10.4018/jaec.2010010103 International Journal of Applied Evolutionary Computation, 1(1), 36-56, January-March 2010 37 Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited. number of international conferences that focus on “autonomics” or “autonomically inspired” topics has increased rapidly in recent years, and the voice advocating standard organizations to make the movement towards the development of interoperability standards for autonomic systems are getting louder. It is evident that this surge of enthusiasm towards AutoComm will accelerate as more industrial companies become involved and relevant standards activities start. The ultimate goal of autonomic communications is to “achieve the autonomy of communication networks with minimum human administration” (Gu et al., 2008), and the self-* capabilities of AutoComm systems are the key to achieving this goal. Gu, et al. identified the principal characteristics of an AutoComm system that distinguish it from conventional communication systems in (Gu et al., 2008). Dobson, et al., summarised the main challenges for AutoComm and reviewed the related work towards tackling some of the challenges in their survey paper (Dobson et al., 2006). In (Raymer, Meer & Strassner, 2008), researchers from the Architecture Expert Group of the Autonomous Communication Forum presented their design of the architecture of autonomic network system. In this architecture, based on the knowledge about the local network situation that is constantly being monitored, the Autonomic Management (AM) element produces and rectifies the network management policies according to the high level business targets, and the Autonomic Control (AC) mechanism proactively configures and adjusts the operations of the network devices according to the latest technical policies. It is clear from many recent literature that, most of the current research on AutoComm has been focused on adding the autonomic management and autonomic control mechanisms into the existing policy based network management (PBNM) mechanism (Gu et al., 2008; Jennings et al., 2007; Raymer et al., 2008). The principle of AC element is straightforward. It monitors the local network’s situation, and feeds the observed data to the AM and PBNM elements. Taking the observed network situational data and the constraints from the PBNM as the reference target, the AC element applies the adaptive control theories to adjust the Policy Enforcement Point (e.g. a server or a router). This is illustrated in Figure 1. The challenge is, when the AM changes its technical policies as reacting to the variation of the network environment and the PBNM accordingly translates the policies into target constraints, the AC has to find an optimal way to adjust the attainable network resources to meet the changing constraints. In Gu et al. (2008), Jennings et al. (2007), and Raymer et al. (2008), traditional control theories have been proposed as potential solutions for this problem. However, these studies assumed that the systems being controlled could be represented by linear models. In the complicated network environment, this assumption cannot always be held because network problems often involve complex nonlinear dynamics, chaotic disturbances and randomness that traditional algorithms are unable to conquer. Genetic Algorithms (GA) (Miettinen & Neittaanmaki, 1999) has been successfully applied to complex optimization problems like adaptive control, optimal control problems, wire routing, scheduling etc. In this article, we adopt GA as an effective tool to address the complex control problem in AutoComm systems. Rather than focusing on adjusting the configuration of network devices (Gu et al., 2008), and services such as web services (Lu, Abdelzaher, Stankovic & Son, 2001), or queue management for TCP-based streaming traffic (Hadjadj, Mehaoua & Skianis, 2007), we take a completely different approach in this article. We focus on the use of GA to control and optimize the communication protocol between two network nodes under the constraints imposed by the PBNM. More specifically, we have designed an evolvable protocol which can be controlled and optimized by the AC and GA in order to equip the communication peers with better autonomy and intelligence. Self-modifying protocols (SMP) are protocols that can be modified at run time by the 19 more pages are available in the full version of this document, which may be purchased using the "Add to Cart" button on the product's webpage: www.igi-global.com/article/self-evolvable-protocol-designusing/40903?camid=4v1 This title is available in InfoSci-Journals, InfoSci-Journal Disciplines Computer Science, Security, and Information Technology. Recommend this product to your librarian: www.igi-global.com/e-resources/libraryrecommendation/?id=2