1 A Swarm Intelligence Based Routing Protocol for Decentralised Cognitive Mobile Radio Networks Andrew Portelli Dept. of Computer and Communications Engineering, Faculty of ICT Adrian Muscat Dept. of Computer and Communications Engineering, Faculty of ICT Abstract Mobile radio networks are renowned for the innate high degree of flexibility they provide to the end user. This is partly attributed to the fact that unlike wire- line networks they do not require any existing infrastructure or central administration. However flexibility comes at a cost, and the biggest challenge in these kind of networks is to find an efficient path between end-to-end communication nodes which aggravate the network throughput when they become mobile or behave erratically. This paper presents an innovative routing algorithm for mobile radio networks that instils an element of cognition at the mobile node itself. The packet transmission protocol, which is based on the ant swarm intelligence meta heuristic, makes the transmission process highly efficient, adaptive and scalable with an increasing number of mobile nodes. It also contributes towards a reduction in routing packet overheads. Index Terms Ad-hoc Networks, Swarm Intelligence, Routing 1. Introduction The ever increasing number of wireless mobile nodes is rendering the management of radio network infrastructures more complex with respect to real time interventions aimed at addressing haphazard network issues. Notwithstanding the multitude of existing research projects in this field focusing on decentralisation, programmable and adaptive networks with the aim of replacing human administration are still far from becoming a reality [4]. Their realisation calls for networks that are aware of their state or needs, have clear knowledge of their goals and ways to achieve them through independent rational decisions and actions. Current network technologies are reactive, in that they tend to adapt themselves by responding to changes in the environment as a consequence of an occurring problem. To cater for the forecasted high increase in wireless mobile radio users in the near future, these networks should evolve to exhibit cognitive characteristics where goals are achieved through autonomous reasoning, adaptive functionality and self-manageability. This paper presents a hybrid on-demand adaptation approach that incorporates a degree of cognition into each node within the radio network, and which actively influences the network when the environment changes. In purely proactive protocols like Destination Sequenced Distance Vector Routing [6] nodes try to maintain at all times routes to all other nodes. Keeping track of all topology changes can become a difficult task especially with increasing number of nodes which are very mobile. Reactive protocols Dynamic Source Routing [7] and Ad-hoc On-Demand Distance Vector routing [8] are in general more scalable. In these protocols, nodes only gather routing information on demand. Before nodes transmit data to a known destination they construct a path, and only when the path becomes infeasible they search a new path. This approach helps in the reduction of the routing overhead. However networks with reactive routing protocols can experience significant drops in performance since these are never prepared for disruptive events. In the presented hybrid approach, nodes have a dual role and embrace the added functionality of a bridge router to forward packets and network status information to other mobile nodes. The implemented algorithm helps the overall decentralised network perceive current network conditions, and as a result plan, decide and act on those conditions whilst taking into account end-to-end goals. The time varying topology of an ad-hoc network makes efficient route selection a formidable task. The problem of simulating mobile nodes has been under investigation for over several years and from various presented approaches aimed at addressing the mobility problem realistically, there seem to be no routing algorithm that encompasses all the characteristics of a true mobile radio network environment [2]. This paper presents an innovative approach for an ad-hoc routing algorithm based on swarm intelligence, a form of artificial intelligence based on the collective behaviour of decentralized, self-organized systems. Swarm intelligence is gaining significant attention in the development of mobile network simulators that can truly help address many of the shortcomings of present network simulation platforms.