کنترل کننده سامانهٔ چند عاملی (MASC) بر پایه منطق فازی

ABSTRACT A fuzzy logic based multi-agents controller

The phenomenal growth and usage of ubiquitous computer network is fueling researches on economic, easy and secured means of connecting to geographically distributed data either in an intranet or on the Internet. The mobile agent technology has been seen as a technology that offers the omnipotent solution to this quest of computer network end-users (Kotz & Gray, 1999). Mobile agents are programs that can autonomously migrate from node to node in a networked environment to perform some computational tasks on behalf of the user. The running code suspends it execution and it is transported to another node or place depending on the destination, where it resume execution from where it stopped. Mobile agents are well known technology which is very effective for implementing complex software and network-centric applications such as e-commerce, telecommunication, Information retrieval systems, process control, etc. The technology offers reduce network latency; encourage optimal bandwidth usage which is be limited by several technical factors. Many end-users (mostly the developing nations’ end-users) still get connected by means of modem, or at most, ADSL over the traditional copper wire. Many other end-users will connect via low-bandwidth wireless networks who cannot afford to have more than 128 Kbps to 1 Mbps available at their desktop or palmtop (Amitava, 1999). The infrastructure and network resources limitations make mobile agent a de facto technology for Internet and network-centric applications. It provides a distinct, common framework in which geographically dispersed, information-oriented applications can be implemented efficiently and easily, with the programming burden spread evenly across information, middleware, and client providers. In other words, mobile agent gives network providers the time and flexibility to provide their users with more useful applications, each with more useful features.

Mobile agent technology has gained a wide range of acceptance among Computer Science researchers and network-centric applications developers due to some salient advantages the technology promise to offer over the traditional client-server architecture. The wide acceptance of the technology is due to its fault-tolerance during network downtime and reduced latency (bandwidth optimum usage) on computer network (Kotz & Gray, 1999). The advantages and potential of agent-to-agent communication has greatly fuel Multi-Agents Systems (MASs) researches, therefore MASs has been seen as a tool to solve many complex Internet applications (Jennings, 2001; Wooldridge, 2002). The authors of Abrahams and Dai (2005) and Espinasse, Fournier, and Freitas, 2008 proposed MASs architecture for information storage, search and retrieval on the web. In the same vain (Mabry, Eetters, Schneringer, Edwards, & Hug, 2003; Sumit, Qutaiba, Jerry, & Aivars, 1998; Webber & Silva, 2008) aimed at solving complex problems of distributed diagnosis, patients monitoring over the Internet through MASs. According to (Mandureira et al., 2008), the appropriate technology for removing the complexity in network scheduling, especially in manufacturing companies’ computer network is through MASs applications. The proposal to solve complex network applications through MASs by researchers has increased greatly with little or no considerations of how the MASs will on bandwidth during their itinerancy. It is clear that, mobile agents are program codes that make use of bandwidth during their itinerancy on computer network. The amount of bandwidth usage is a function of the number of agents in the systems (Baek et al., 2001a, 2001b) and our simulation experiment support this view. To strike a balance between the numbers of agents in a MASs and the available bandwidth this paper propose a flexible fuzzy logic based multi-agent controller that will regulate the number of agents release to the network based on available bandwidth. The rest of the paper is organize in the following way, Section 2 elucidates some FL controllers, Section 3 describes our model in FL environment, Section 4 discusses our simulation environment and simulation results while Section 5 gives the conclusion of the paper.