The Research On Coordination Of The Task On The Internet Of Things Based On GPGP Mechanism

With the rapid development of wireless communication technology, RFIDtechnology, sensor technology, intelligent control technology, technology, significantchanges to the existing network will be happen, it will be proposed to the Internet ofthings networking concepts, and it has become bring the academic research focus.At present, the development of the Internet of things have to face manyproblems. First, Internet of things involved in the collaboration of a wide range ofdevices to complete a task at the same time, these devices in the same task groupwork together to accomplish the same goal, then make a formation of the complexdependencies. In order to solve these problems, Agent technology and middlewaretechnology has been brought into the Internet of Things to solve these problems.Within Multi-Agent domain, multi-agent’s coordination is an important researchquestion, the existing coordination technology, however, cannot adapt to the zeroInfrastructure characteristics of the Internet of Things. Unlike traditional network,the devices in the Internet of Things are usually dynamically join the network, andthese devices may come from different manufacturers, be complied with differentformats and have different ability, so the coordination mechanisms in the Internet ofThings must be able to handle this problem, it must transfer the information ofcoordinate more far.(1)In order to solve the multi-agent coordination’s problem of the zeroinfrastructure. We make a improvement for the traditional GPGP Multi-Agentcoordination mechanisms based on graph theory, so that the information ofcoordination between different agents can transfer in a more a wider range. In thisarticle, by improving the process of “detection coordination” in traditional GPGPmechanisms, and add the additional information into the process, so the improvedalgorithm needs to deliver the message in a wider range, but in most Agent withinthe same broadcast domain, it will not bring too much messages increase.(2) If the Agent distribute at a wide space, there’re more agents in the task group can’t communicate directly, this will bring a significant increase in the numberof messages to coordinate the network Therefore, we proposed a clustering method,and we divided the agents into different clusters, using the improved coordinationmechanisms within the cluster and using the traditional mechanism betweendifferent clusters, thereby reducing the communication cost.This paper use the simulation tools--repast to compare the improvedcoordination mechanisms and traditional mechanisms in different environments, andmake the analysis concluded. And make experiments on most of the agent within thesame physical communication range, and without clustering mechanism.