On the patterns of the nonstationary datagram based fast communication processes

Abstract

. Nowadays expectations against modern communication services involve not just Quality of Service (QoS) enhancement for real-time applications but also increased transmission rate between the storing and processing of Big Data nodes. Transmission Control Protocol (TCP) has strict flow con-trol of the data stream providing automatic adaptation to the path load of the process-to-process communication. User Datagram Protocol (UDP) based solutions are proposed to settle the communication efficiency. In this paper, we analyse the effect of three independent communication parameters on the efficiency of looped UDP communication: the size of the Maximum Transfer Unit (MTU), the bandwidth of the end-to-end session, and the segment size of the UDP protocol data unit. The usage of nonstationary multi-resolution methods helps to identify three characteristic patterns offering identification of the objective qualitative features of the looped datagram communication services.