Presentation on theme: "6/14/20141 A Cluster Formation Algorithm with Self-Adaptive Population for Wireless Sensor Networks Luis J. Gonzalez."— Presentation transcript:
6/14/20141 A Cluster Formation Algorithm with Self-Adaptive Population for Wireless Sensor Networks Luis J. Gonzalez
6/14/20142 Abstract Discussion of the correlation between the population size and performance of artificial social insect colonies along with the potential use of self-adaptive techniques to overcome the limitations of using parameters with manual predetermined values in cluster formation algorithms for wireless sensor networks (WSN).
6/14/20143 Eusociality Eusociality, the division of labor without any known centralized leadership, and the effectiveness to find the shortest path between the nest and a food source are characteristics of many insect societies.
6/14/20144 Artificial Insect Colonies Population Constraints The success of eusocial insect colonies is based on castes with specialized reproductive and non-reproductive functions. The level of specialization and population size growth is a natural response to the stimulus created by the ecological context. However, this self- adaptation to the environment cannot be achieved in artificial insect colonies where the population size parameter is manually predetermined.
6/14/20145 Castes, Pheromone, and Encounter Rates Eusocial insects are morphologically different and divided into castes depending on their functions within the colony. Ant, honey bees, and termite colonies are integrated by reproductive and worker individuals.
6/14/20146 Queens, Drones and Workers The queens are the essence of the colony because they produce the eggs. The drones, as the queen's counterparts; thus, queens and drones are the starting point for the endurance of the colony. All the workers are female and traditionally perform non-reproductive functions, they can be patrollers, foragers, breeders, or responsible for the nest maintenance.
6/14/20147 Selfish Behavior There are some cases where workers reproduce selfishly rather than performing their traditional non-reproductive duties, which may impacts negatively the performance of the colony.
6/14/20148 The Stimulus- Response Pattern The survival of a colony depends on the cooperative natural intended work of their members. The selfish behavior of workers alters the population size and the natural operation of the colony. The population size may be increased when the performance of the colony is altered by the selfish behavior of workers. An optimal population size is required to balance opposing selection pressures.
6/14/20149 Artificial Insect Colonies An evolutionary approach is used to reach a level of specialization for the artificial workers. Thus, a set of parameters, i.e., population and number of trials, must be defined to reach an optimal solution. However, the values of these parameters are not necessarily optimal when they are calculated manually.
6/14/ Predetermined Population Values In the case of artificial insect colonies, the population cannot respond to the ecological stimulus because the values are pre- determined.
6/14/ Self-adaptation The population size and level of specialization of workers are fundamental for the efficiency of artificial insect colonies; therefore, the parameters that control those variables should be the response to ecological stimulus, which can be achieved if self-adaptive techniques are considered.
6/14/ A Cluster Formation Algorithm with Self- Adaptive Population Wireless sensor networks (WSN) can be defined as a set of small spatially distributed autonomous battery powered devices or sensors, which interact and cooperate between them to monitor physical or environmental conditions. The efficacy of WSN depends on the minimization of package collisions, control packet overhead, and overhearing of unnecessary traffic and idle listening to avoid energy wastage, which is the scarcest resource in WSN.
6/14/ Minimize Energy Consumption The formation of clusters with greater affinity to the cluster leader helps to optimize package transmission and reception, and minimize energy consumption.
6/14/ Hypothesis The level of specialization or cluster's efficiency depends on the cluster size previously predetermined; however, the cluster sizes are not necessarily optimal when they are calculated manually. The use of biologically-inspired self- adaptive techniques to set the cluster size can maximize the formation of a uniform population of several clusters with greater affinity to the cluster leader, which will reduce the energy wastage.
6/14/ Architecture and Operation of a wireless sensor network The operation of WSN encompasses the cluster and sink tree formation phases.
6/14/ Optimal Cluster Size Calculation Having a few clusters, which can be counted by the number of leaders or cluster heads, will overload the cluster processing capacity. On the contrary, too many leaders with few sensors will cause idleness or under use of that node; hence, an optimal cluster size is essential for load distribution in WSN.
6/14/ Standard Deviation The standard deviation of cluster sizes can be used as an indicator to determine the optimal cluster size because "the average cluster size is inversely proportional to the average number of clusters"
6/14/ Standard Deviation The calculation of the optimal cluster size will have to be oriented to obtain the smallest standard deviation in the WSN. The smallest standard deviation suggests that the load is distributed uniformly among the leaders or cluster heads. The optimal cluster size will also help to minimize the load inequality and extend the overall system lifetime.
6/14/ Conclusion The success of eusocial colonies is based on the division of labor according to reproductive and non-reproductive worker castes. The growth of the non-reproductive castes (the workers) depends on the ecological context; the population size is naturally adjusted according to ecological demands. Self-adapting the population size in the cluster formation may contribute to the creation of energy efficiency wireless sensor networks.