Research On Improved RED Algorithm For Internet Of Things Network Congestion Control

With the ever-changing of network technology and social development needs, internet of things as an emerging pattern of rapid rise in the past few years, and has been rapid development, though the concept is just emerging, but in the next few years it will be in a very change people’s way of life to a large extent on the production. At the same time we should also note that people will put forward higher requirements, including the transmission of network transmission speeds higher performance and better network quality of network services.So, internet of things to bring convenience, but also gives us a new research topic-Things network congestion control.Aiming internet of things pattern complex network environment, first of all, in the analysis of random early detection algorithm based on the principle RED, is proposed based on the upper threshold of RED improved algorithm--MARED.MARED algorithm improved original RED algorithm average queue length and the packet loss probability function curve,depending on the queue length different drop probabilities to accommodate different network congestion control state.Compared to the RED algorithm major improvements have been made to the following two aspects:1.Through the analysis of the average queue length shows that the router buffer queue length distribution is non-linear, and is a decreasing exponential distribution, this article by the linear relationship of the RED algorithm and packet loss rate was changed to queue in long lines curve, with a more realistic situation, to achieve a smooth transition of packet loss rate, to ensure that in any case can be made for fast and accurate response to congestion.2.Because RED algorithm queue length reaches the maximum value maxth door directly into a packet loss rate, discarding of all newly arriving packets, sets a new parameter- the upper threshold Uth. When the queue length maxth between Uth, MARED algorithm maxp loss rate gradually increased from 1, thus improving the utilization of the link bandwidth, and can provide greater data throughput and the packet loss rate is smaller.Secondly, the improved algorithm stability analysis. Systems using TCP fluid model,using 2-D Hurwitz-Schur stability determination principles, combined with this paper, the improved algorithm MARED its loss strategies are analyzed and presented stable operation of the algorithm parameters area.Finally, the performance of this algorithm simulation. By installing a network simulationtools in a virtual machine NS2, in a simulation model of things in a mixed network environment simulation data obtained show that the improved algorithm of this paper MARED average queue length, data throughput and packet loss rate, etc. their performance is better than the original RED algorithm. Therefore concluded that the proposed algorithm can adapt to the complex network environment of things, to better meet the needs of practical application.