Ant Colony Optimization and Swarm Intelligence

Swarm intelligence is an emerging field of biologically-inspired artificial intelligence based on the behavioural models of social insects such as ants, bees, wasps and termites. Respective algorithms are made based on ants, bees, wasps such as Ant Colony Optimization (ACO), Particle Swarm Optimization (PSO) and Bee Swarm Optimization (BSO). It is a scientific theory about how Complex and sophisticate behaviours can emerge from social creature group. It helps in designing a framework for designing distributed algorithms which are originally derived by studying models of social insect behaviour. Swarm intelligence is a probabilistic method for building probabilistic paths between nodes based on simple rules.

In general, ant colony optimization meta-heuristic tries to solve a combinatorial problem using the collaboration of a group of simple agents called artificial ants. ACO routing algorithms establish optimum paths to the destination using a number of artificial ants that communicate indirectly with each other by Stigmergy. Stigmergy is a way of indirect communication between individuals which, in an ad hoc network case, is done through the modification of some parameters in the nodes of the network.

The Ant colony optimization is based on the foraging behaviour of ants. When ants search for food, they wander randomly and upon finding food return to their colony while laying a chemical substance called pheromone. Many ants may travel through different routes to the same food source. The ants, which travel the shortest path, reinforce the path with more pheromone that aids other ants to follow. Subsequently more ants are attracted by this pheromone trail, which reinforces the path even more. This autocatalytic behaviour quickly identifies the shortest path.

Over the years, the development of energy-efficient communication and networking algorithms and protocols for MANETs has been a great challenge. With the ubiquity of wireless communications and the advent of new wireless technologies, network communication has been revolutionized, enabling new techniques to deal with low power and energy efficiency in MANETs. Due to their infrastructure-less nature, MANETs require the use of multiple hops for connecting all the nodes to each other. Consequently, the relaying of the messages from one mobile node to another, and the peculiarity of the wireless transmission medium (i.e. noise, interference, etc) are some of the major issues that can be raised. In addition, when the power source of a node in MANET is low or costly, the energy efficiency becomes a key concern, as it is directly related to the design and operation of such networks, particularly the routing operation.

Routing in MANETs takes place through mutual communications. Due to the lack of infrastructure, various routing algorithms, that include 2-way communication between the nodes for route selection purposes, are used. Each node needs to be equipped with specific configuration which helps the node in route discovery. This involves information regarding the current states of other nodes, information on the links that the node has with other nodes, a routing algorithm that it needs to follow for packet routing. The challenge involved in routing packets in MANETs is the same. Each node needs to be constantly updated with the current network state and active/inactive nodes, and doing on-the-fly requires a routing algorithm. 

There are different routing algorithms focusing on different optimization parameters such as time delay, path length, bandwidth, maximum link capacity, among others. This paper proposes an ACO-based energy-efficient routing algorithm for packet routing in MANETs (so-called ACO-EEAODR), which takes the battery powers of nodes as the primary criterion for route selection, then optimizes the network performance by finding the energy-efficient routes. ACO is a learning algorithm, where information regarding past links and their worthiness are stored and used to find an optimal route.