Message Ferry Route Calculation through K-means Clustering Algorithm for Partially-Connected MANET

Partially-connected MANETs are the Wi-Fi networks where most of the time a comprehensive path between source and destination does not exist due to narrow radio transmission range of the nodes, low density of nodes, wide deployment area, physical obstacles like high rise buildings, severe weather conditions, or other physical factors. The traditional routing protocols of the Mobile Ad-hoc Network presume that the network is connected. There are two possible solutions to solve the connectivity problem in sparse MANET. The first solution makes use of random movement of nodes to transfer messages whereas in the second, one node is given special privileges to roam in the deployment area to distribute messages in segmented areas, which is also called the Message Ferry (MF) approach for transmitting data. Message ferries keep moving in the deployment area and take the charge of storing, carrying and forwarding data of the regular nodes. In the present paper, the ferry-initiated approach has been adopted with 2 ferry nodes, which are allowed to move in such a way that they cover the maximum area of the grid. It is proposed that the ferry nodes will design their routes dynamically on the basis of current topology of MANET and in one round trip it will pass through the centers of the partitioned clusters formed by the regular nodes. The process of calculating cluster center is repeated in further trips. K-means clustering algorithm is applied to calculate the cluster centers of the partitioned clusters. The results indicate positive performances of the proposed model in various simulation experiments conducted.

Solving Closed Walk Ferry Route Design Problem by Using an Improved Genetic Algorithm

In recent years, using message ferries as mechanical carriers of data has been shown to be an effective way to collect information in wireless sensor networks. Closed Walk Ferry Route Design (CWFRD) problem aims to minimize the average weighted delay of the sensed data to the sink by designing the ferry route as a closed walk which contains more than one simple cycle. In this paper, an improved genetic algorithm is proposed to solve the CWFRD problem. Different to other problems solved by the genetic algorithm, in the CWFRD problem, the ferry may access the sink node more than one time, and then the lengths of the two chromosomes for evolving may be different. The proposed Improved Genetic Algorithm (IGA) can solve this problem, and experimental results show that IGA can greatly reduce the average weighted delay.