Research On Transmission Performance Optimization In Enterprise Wireless Networks

With the development of communication and integrated circuit technology,Wi-Fibased wireless networks have become an indispensable part of social production and life practices.According to Cisco's measured data and forecast data,the proportion of Wi-Fi traffic in all IP traffic will increase from 43% in 2017 to 51% in 2022.More than 79%of the data depends on the latency and throughput performance of data transmissions.A 20-millisecond delay between the client device and the access point in enterprise wireless network will cause the page-loading time to be magnified to 3 seconds,and this page-loading time exceeds 47% of users' expectations.Researchers of industry and academia have proposed many methods such as opportunistic scheduling,load balancing,transmission protocol,and optimization to improve the delay and throughput performance of data transmission,but the problem is still serious.Enterprise wireless network,which acts as the most important Wi-Fi wireless network organization form in the practice of production and life,also has the above problems.Still,enterprise wireless networks have unified management,software-defined,can support complex computing,provide a more comprehensive network status and information,and they can achieve better latency and throughput performance.Researchers of software-defined enterprise wireless networks can focus their works on exploring network status and focusing on how to design and evaluate network optimization algorithms without being restricted by device compatibility,time-complexity,message communications,and other issues.We utilize these advantages of the software-defined enterprise wireless network,and focus our works on opportunistic scheduling,load balancing,and transmission protocol optimization to optimize the delay and throughput performance of data transmissions.We also implement and evaluate our algorithms on our platform.The contents of our researches are as the following.(1)We study the quantitative relationships of Wi-Fi characteristic variables on data transmission congestions.Our research works show that congestions do have significant relationships with some of the characteristic variables,based on these prediction relationships,we implement and evaluate a TCP congestion prediction and avoidance method,and the experiment results show that this method has noticeable data transmission performance enhancement on both delay and throughput.(2)We study the handoff-delay-oriented load-balancing problems in enterprise wireless networks.The handover delay commonly exists in practical wireless networks,but its impacts are not well studied.We model the impacts of handover delay,and formulate the load-balancing problem as a utility maximization problem.We prove that this problem is NP-hard.We then relax this problem to a discrete linear problem and get the fractional solutions,and then we round the fractional solutions to 0-1 integer solutions.We give the corresponding algorithm and compute its approximate ratio.We implement and evaluate the algorithm on our platform.The experiment results show that our algorithm has noticeable data transmission performance enhancement on both delay and throughput.(3)We study the delay-constraint transmission scheduling problem in enterprise wireless networks.We model this problem as a delay-based utility-maximizing problem.We design virtual queues and transform the original problem into a queue-stabilizing problem,and then we design a queue-stabilizing problem based on Lyapunov theories,we prove that our algorithm is throughput optimal and it guarantees the delay constraints.Then we evaluate our algorithm's theoretical performance;we also implement our algorithm on our platform and evaluate its practical performance,and the experiment results show that our algorithm has noticeable data transmission performance enhancement on both delay and throughput.(4)We study the transmission control protocol optimization in enterprise wireless networks.Our proxy-based optimization method based on cooperative game theories aims to optimize the delay and throughput performance of TCP data transmission by utilizing the abundant information on access points.We model the TCP data rate control problem in enterprise wireless network as a utility optimization problem.As TCP protocols are canonical coalitional game methods,we utilize the canonical coalitional game method,differential non-cooperative game method,and coalition-forming game method to solve this problem individually.We prove that the canonical coalitional game method has the best throughput performance.Then,based on the gradient of payment to the rate,we propose a rate control algorithm to enhance the TCP performance.The simulation results show that our algorithm does improve throughput performance.