Low-delay Transmission Design For Instant Messaging Service

Instant Messaging(IM)is one of the most popular and important services of the internet,especially the mobile internet,it is essential to satisfy the needs of IM and provide the IM users with good experience for system designers.IM includes two services,i.e.the presence update service and the instant message transfer service.Low-delay transmission is the basic requirements of IM.However,providing low-delay transmission for the two services of IM may face many challenges.In presence update service,small packets are sent periodically,which may interrupt the sleeping of Mobile Devices(MD),leading high energy consumption.Multi-hop transmission is able to reduce the energy consumption,while the high-energy-consumption problem of relay nodes and high latency come along.In the instant message transfer service,IM users are concurrently launching multiple sub-services(video chat,audio chat and text chat),here low delay means that the IM users' sense of latency is low.However,the IM users' sense of latency is affected by many factors,i.e.the latency of sub-services,the users' preference of sub-services and the parameters of networks.This thesis focuses on the design and analysis of the low-delay transmission algorithms for IM.For presence update service,the wireless multi-hop networks provision by energy harvesting are studied,and the relay selection algorithm that provides the minimum end-to-end delay under the energy constraint is proposed.A new data aggregation and forwarding method is proposed to reduce the energy consumption of relay nodes and to alleviate the energy constraint,in which the MDs collect the packets sent to them while they are idle,then,at the time of sending,they repack all the packets and send them all at once.The minimum end-to-end delay problem is formulated as a finite-states Markov decision process to solve the non-causal problem invloved.The"states-optimal relay selection" table is generated offline and the optimal relay selection can be found rapidly by searching the table online.For instant message transfer service,the model of IM users' sense of latency while one IM user concurrently launches multiple sub-services is studied.And the low-delay scheduling algorithm based on the model is designed.The Attention Factor(AF)is defined to reflect the probability of one sub-service being focused by the user.The sub-service that gets the largest AF is defined as the Representative Service(RS),which is used to determine the sub-service that the user is focusing on.The Normalized Score(NS)is defined to provide a unified measure of the latency of different sub-services.The NS of the RS is defined as the Representative Service Quality(RSQ),which is the measure of the users' sense of latency.The AF update process is also designed to trace the sub-service that the user is focusing on.Based on the model,the Minimum Representative Service Quality First(MRSQF)scheduling algorithm is proposed,which is a two-step algorithm.The first step is to find the user with the minimum RSQ.Of all the sub-services that the user launches,one sub-service gets the chance of transmission according to the AFs of the sub-services.The algorithm allocates the scarce resources to the sub-service that most affects the user's sense of latency,reducing the largest delay of the system.