Research On Performance Statistics Of Distributed Queue Random Access For Massive Machine Type Communication

Massive Machine-Type Communication(m MTC)is mainly oriented to the Internet of Things(Io T)with the goal of sensing and data collection.As an important part of Io T and one of the three core application scenarios of the fifth-generation mobile communication system(5G),m MTC aims to realize Internet of Everything through the effective connection billions of machine-type devices.However,when a large number of terminals are connected to the network synchronously,the classic contention avoidance type random access technologies(such as ALOHA or CSMA series)are limited in throughput due to severe access conflicts,and cannot meet the demand for m MTC massive connection;while contention resolution type random access technologies(such as tree splitting series)are expected to achieve a substantial increase in random access performance through the introduction of conflict resolution or scheduling mechanisms,which is an important research direction for future m MTC random access.Distributed Queue(DQ)random access is a variant of the tree splitting protocol,which combines multi-tree splitting algorithm with a set of simple intelligent rules,so that the terminals can execute the protocol rules in a distributed manner.It allows the terminals to know when its own turn to transmit,and indirectly obtain transmission access permission while avoiding conflicts completely.Theoretically,DQ allows unlimited terminals accessing and is stable under any load conditions,which is especially suitable for solving the massive connection problem of m MTC.However,current performance analysis of DQ random access in m MTC is mainly based on numerical simulation or cascaded Poisson queue model.There is no statistical model that completely characterizes the dynamic working process of DQ random access and a complete theory of performance statistical characteristics research framework,which severely restricts the full manifestation and further improvement of its performance advantages.In view of this,the paper proposes a performance statistical analysis method based on sample tree,then analyzes performance influencing factors,and on this basis,further proposes Monte Carlo method based on sample space to approximate performance,all of which can provide reference for the actual deployment and optimization of m MTC.The research content includes:(1)By displaying the dynamic process of DQ access scheduling in a tree structure,we define the access scheduling example tree of a certain batch of arrival as a sample tree,so that establish a sample tree model.Then,by mining the statistical characteristics of the distribution of competing devices in each frame,we construct a complete sample space and its statistical relationship with the sample tree,and derive the probability mass function of the sample tree,which provides theoretical foundation model for the subsequent theoretical analysis of performance statistical characteristics and its extended research.(2)Based on the statistical characteristic relationship of the sample tree and sample space,statistical characteristic analysis method of each performance indicators is given,including the theoretical expressions of the probability density function,mean and variance of throughput,latency and energy consumption.Numerical simulation results verify the accuracy of the proposed analysis method,prove the stability of DQ random access,and demonstrate the influence of the number of machine-type devices,number of contention slots and maximum number of transmissions on the above performance,which provides a reference for exploring the optimal parameter settings of the system to achieve compromise optimization of performance.(3)Aiming at the problem of high algorithm complexity and slow running speed of the performance statistical characteristic analysis method based on sample tree when the number of access devices is large,we propose a Monte Carlo method based on sample space.It first estimates the size of the sample space,and then sets the number of sampling simulations reasonably according to it,so as to approximate the performance.The simulation results show that when the number of devices is large,the algorithm can increase the analysis efficiency by nearly a hundred times while ensuring the accuracy of the estimated performance.