Research On Coding Technology Of Fog Computing Platform For Internet Of Things

In recent years,with the continuous development of the Internet of Things,the access of massive heterogeneous intelligent terminal devices has caused the explosive growth of data in the network.The traditional centralized cloud computing architecture faces many challenges.As an emerging computing paradigm,fog computing is an extension of cloud computing.It extends computing,storage and other resources in the cloud to the edge of the network,and handle most requests from terminal devices in the fog layer.Coded MapReduce is a novel computing framework.Based on the traditional MapReduce framework by redundant storage and calculation in the Map phase and coded wireless multicast in the Shuffle phase,transmission overhead during the Shuffle phase is significantly reduced,and the reduction percentage is positively correlated with the redundancy calculation factor.In view of the fact that the transmission overhead of the related coded MapReduce schemes remain to be improved,in this paper for the application scenario of the fog computing platform for the Internet of Things,in order to make full use of the rich computing resources at the edge of the network the schemes related to coded MapReduce is analyzed and studied.New coded schemes are proposed respectively under two typical network topologies:For the scenario where the fog nodes are connected to each other through the star wireless network topology,a low division number coded MapReduce(LDCMR)scheme is proposed.During the data set allocation phase by strictly limiting the number of segments of the input data set,the number of coded words the amount of data actually transmitted in the Shuffle phase are reduced.For the scenario where the fog nodes are connected through a three-tier tree wireless network topology,the access point decoding hybrid coded MapReduce(ADHCMR)scheme is proposed.Secondary access points instead of the other nodes in the same layer perform decoding in the cross-group shuffling phase to reduce the transmission overhead of the secondary wireless access points during the Shuffle phase.In this paper,the proposed schemes and related existing schemes are theoretically analyzed under the corresponding system models,and the feasibility and correctness of the schemes are verified by simulation.The results show that the LDCMR scheme can effectively reduce the actual amount of data transmitted in the Shuffle phase when the intermediate value density mapping relationship satisfies the characteristics of monotonous increment and convexity.When the redundancy calculation factor is 2,the ADHCMR scheme can further reduce the total transmission overhead of the secondary wireless access points during the Shuffle phase.