According to the limitations of calculation of the original random early detection (RED) algorithm in linear packet loss rate. This paper proposes an improved algorithm which imposes nonlinear smooth for packet loss rate function of RED algorithm. The speed of growth of packet loss rate is relatively slow near the minimum threshold, while near the maximum threshold the speed of growth of packet loss rate is relatively faster. In this case, using the trend of the average queue length to dynamically adjust the parameters of the RED algorithm, it reduces the dependence on the parameters of the RED algorithm and enhances the stability of the algorithm. NS simulation shows that this algorithm has been significantly improved for packet loss rate, throughput and other performance.

No abstract

To solve geometric object crawling issues for visual robot,here we try to place series of specially designed markers in the geometric center of the surface, and store the relevant information of the corresponding object to the identified ID,so that the object crawling issue has become the the location and identification of markers ID.In this paper the author designs a marker system on Hamming code, detects the identity quadrilateral by counterclockwising finishing contour point and calculating the the contour perimeter in the image processing procedure,and then identifies the ID with the the decoding theory. Simultaneously Harris algorithms are chosen to extract the four corner coordinates of the sub-pixel accuracy. Experiment and theory show that the method can achieve the object location and recognition .