Multiagent systems arise in several domains of engineering and it seems that can be also efficiently used to optimise resources in present and future generation of wireless heterogeneous networks. In this context an agent is considered as a software entity with autonomous behaviour and objectives, embedded in an environment which perceives, and in which it acts. Autonomy means the ability to have control over its actions and its internal states, and therefore implies de-centralized control. Key features of mobile multiagent systems are mobility (ubiquity), network awareness, communication, intelligence, reactivity, autonomy, goal oriented, temporally continuous, learning, flexible and character.
So we can summarize that a mobile multiagent system is a set of intelligent mobile agents, interacting and trying to reach a set of goals, which in the case of wireless systems are oriented to an efficient sharing of resources and load balancing between nodes.
On the other side Mobile and Wireless networks (heterogeneous networks) are characterized by being large scale networks with different radio access technologies (GSM/GPRS, UMTS, WLAN, WiMax, Bluetooth, Ad-Hoc,…), different cell structure and equipment (acces points, nodes B, Base Station Transceiver, macrocells, microcells, picocells,..) as well as different user requirements and different services, which change dynamically (the channel is time variant, user mobility, ubiquous connectivity in ad-hoc networks, etc.). This kind of networks require sophisticated optimization-based techniques and negotiation techniques to efficiently manage all the resources and parameters.
When advanced services are implemented in a wireless mobile network, more complex management algorithms are required, and what is crucial, they should be both fast and reconfigurable. The use of Mobile Agents could be very efficient in preserving user requirements while maintaining Quality of Service, by dynamically optimizing the heterogeneous network and all the resource management algorithms (RRM and CRRM) . Routing protocols to handle inter-working of ad hoc mobile networks and infrastructure mobile networks could be optimized.
To start Diploma Thesis activities it seems more feasible if initially multiagent is applied to ad-hoc networks, due to the decentralized structure of this type of network. Agents operating in an ad-hoc network environment must reason about complex network information that can include dynamically changing the topology, resource-constrained computing devices, multihop routes, signal strength, packet loss, jitter and bandwidth limitations. It is expected that they contribute to reduce and optimize network traffic, while balancing the traffic among nodes, so increasing the performance and capacity of the network nodes. Some examples of applications of ad-hoc networks are sensor networks, disaster relief, personal area networks, embedded computing, conferencing, among others.
After learning and doing an exhaustive analysis of performance over ad-hoc networks, some rules and protocols could be exported to infrastructure wireless networks (as WLAN and WiMax) and finally to mobile networks (GSM/GPRS and UMTS networks). |