Genetic Algorithm Based Resource Minimization in Network Code Based Peer-to-Peer Network
Block scheduling is difficult to implement in P2P network since there is no central coordinator. This problem can be solved by employing network coding technique which allows intermediate nodes to perform the coding operation instead of conventional store and forward the received data. There is a general assumption in this area of research so far that a target download rate is always attainable at every peer as long as coding operation is performed at all the nodes in the network. An interesting study is made that a maximum download rate can be attained by performing the coding operation at relatively small portion of the network. The problem of finding the minimal set of node to perform the coding operation and links to carry the coded data is called as a network code minimization problem (NCMP). It is proved to be NP hard problem. It can be solved using genetic algorithm (GA) because GA can be used to solve the diverse NP hard problem. A new NCMP model is proposed which considers both minimize the resources needed to perform coding operation and dynamic change in network topology due to disconnection. Based on this new NCMP model, an effective and novel GA is proposed by implementing problem specific GA operators into the evolutionary process. There is an attempt to implement the different compositions and several options of GA elements which worked well in many other problems and pick the one that works best for this resource minimization problem. Our simulation results prove that the proposed system outperforms the random selection and coding at all possible node mechanisms in terms of both download time and system throughput.