Kozo Satoda

Constructing stochastic model of TCP throughput on basis of stationarity analysis

End-to-end TCP throughput on best-effort networks fluctuates widely depending on cross traffic and radio interference, which degrades the quality of public cloud services. However, if throughput variation is predictable on the application layer, service providers can be proactive in solving the problem of quality deterioration. We argue that TCP throughput fluctuation occurs in two states, stationary and non-stationary, and developed a method for characterizing these states. We also constructed a stochastic model of TCP throughput, which is a mixture of a stationary process model and a non-stationary one based on stationarity analysis. An evaluation experiment showed that our proposed model could predict the next 200 seconds of TCP throughput fluctuation through the Internet on three types of access networks (HSDPA/LTE/Wi-Fi) with an average accuracy of about 90%.

Adaptive video pacing method based on the prediction of stochastic TCP throughput

Many of the video streaming services on the Internet and mobile networks use HTTP/TCP for video transport. Video streaming clients are generally designed to download a sufficient amount of data to avoid playback interruption caused by the lower TCP throughput. Storing more video data than necessary may generate a large amount of wasted traffic when users quit playing video before all of the downloaded video data are played. To prevent downloading unnecessary video data, HTTP downloading with just-in-time delivery features, called video pacing, is focused on. However, video pacing may cause playback discontinuity, resulting from a decrease in the playout buffer size. This is because conventional video pacing methods control the playout buffer size without considering the TCP throughput fluctuation. In this paper, we propose an adaptive video pacing method that enables both reducing unnecessary video traffic and maintaining user-perceived quality. The proposed method has two unique video-pacing functionalities, namely, predicting the stochastic diffusion of TCP throughput and controlling the playout buffer size by taking into account the future throughput. Experimental results from simulating mobile networks show that the proposed method can decrease the average playout buffer size in stable environments by up to 42.62% compared with the conventional methods. Moreover, the frequency of the playback discontinuity of the proposed method is shown to be up to 34.1% less than that of the conventional methods in unstable environments.

Proposal and evaluation of Joint Rate Control for stored video streaming

Rate control, which involves adapting the sending rate and video rate to network environments, is effective for stably streaming stored video over unmanaged IP networks. In this paper, we propose a novel rate control technique for optimizing the evaluation function defined as user-perceived video quality using an optimal control law in modern control theory. We call our method Joint Rate Control because it integrally controls both the sending rate and video rate. The results of numerical evaluation show that Joint Rate Control improved the evaluation function value by about 20%–50% from that of a conventional rate control method. We have demonstrated that Joint Rate Control optimizes the video quality; that is, this control method was the most effective in streaming video.

Quality evaluation of voice over multiple TCP connections

Voice communications over the Internet are growing rapidly thanks to faster Internet and mobile networks. Although multimedia communication techniques such as voice over IP (VoIP) have been using UDP, TCP is gaining gradual popularity since it enables passage through network address translators (NATs) and firewalls. Many researches have been done on real-time communications using TCP, which indicates the possibility of achieving low delay communications over TCP. However, little mention has been made of user-perceived voice quality over TCP. This paper presents a VoIP system using multiple TCP connections. We conducted thorough voice quality evaluations over these connections by changing network and VoIP application parameters. Our evaluations show that voice communications can be achieved in 10% packet loss rate and 100 ms round-trip time environments if the number of TCP connections, the sending unit and the decoding buffer size are optimized. We also provide optimal parameter guidelines for achieving high quality voice communications over multiple TCP connections.

QoE maximizing bitrate control for live video streaming on a mobile uplink

Video streaming is now one of the main applications on mobile networks. Since the throughput fluctuates widely on a mobile network, adaptive bitrate control methods that control the video bitrate in accordance with this fluctuation have been developed. These methods decide the bitrate at a one-second or a few-second interval. In most of these methods, the output of a decision is a bitrate value, and the decided bitrate continues until the next decision. This paper models the behavior of video packets on a mobile uplink and proposes a novel bitrate control method. In the method, the output of a decision is the bitrate change over time for several seconds into the future, not a bitrate value, that maximizes an integrated value of the quality of experience (QoE) indicator for the model of behavior. This paper compares the proposed method with conventional methods by simulation. The proposed method improved the mean opinion score (MOS), which indicates the estimated QoE as a value between 1 and 5, by up to 0.2.

Autonomous off-peak data transfer by passively estimating overall LTE cell load

A method for off-peak transfer of data, by which user equipment (UE) autonomously shifts a peak load on a base station, is proposed and evaluated. In regard to the proposed method, how UE monitors the load on a base station is a significant problem. To solve this problem, a novel method of passively estimating the resource block (RB) usage rate of a serving cell on the basis of radio conditions measured by UE is presented. The estimated RB usage rate can be regarded as the traffic load on the base station. The effectiveness of the proposed methods for load monitoring and off-peak transfer of data was validated through experiments on commercial LTE networks.

Robust available bandwidth estimation against dynamic behavior of packet scheduler in operational LTE networks

We shed new light on the mechanism behind how the dynamic behavior of a packet scheduler at the link layer in mobile networks degrades the accuracy of conventional available bandwidth (i.e., unused capacity of an end-to-end path) estimation methods that use a probing packet train (i.e., a set of multiple probing packets). Most of the conventional methods, which were originally designed for wired networks, estimate available bandwidth at the receiver by detecting changes of the observed queuing delays of probing packets. They utilize a microscopic approach in which they check the difference of the queuing delay of each packet on a packet-by-packet basis in order to detect the queuing delay changes. We found that the dynamic behavior of a packet scheduler at the link layer dramatically disturbs the queuing delays observed at the receiver. The disturbed queuing delays make it tremendously difficult for the conventional microscopic approach to detect changes of the delays, resulting in degraded estimation accuracy.

A QoE indicator and a transmission control method for VoIP on mobile networks considering delay spikes

Voice over IP (VoIP) services are now offered on many mobile networks. However, delay spikes, which are sudden large increases in packet delay, affect the quality of VoIP on mobile networks. This paper proposes a quality of experience (QoE) indicator considering delay spikes. The QoE indicator combines a voice quality indicator such as ITU-T P.862 and the delay part of ITU-T G.107, which is a quality indicator for VoIP. This paper also proposes a transmission control method that selects the codec type and the sending interval of voice packets so as to maximize the QoE indicator. The method improves the mean opinion score (MOS), which indicates QoE as a value between 1 and 5, by up to 0.4.

Study on mechanism and reduction approaches of delay spikes occurrence on mobile networks

Since smartphones have spread and the transmission rate of mobile networks has accelerated, a lot of VoIP services are provided on mobile networks. Voice quality degradation due to delay spikes (i.e. sudden large increases in the packet delay) is the key issue for VoIP on mobile networks. To resolve this issue, adaptive jitter buffer control methods have been proposed to deal with delay spikes. These methods control the length of the jitter buffer in order to reduce voice quality degradation due to delay spikes. For further improving the quality of voice communication, it is effective to reduce the number of delay spikes. This paper analyzes the mechanism of delay spikes occurrence and proposes delay spike reduction approaches by controlling size and interval of voice packets on the basis of radio quality. Simulation results on the network simulator (ns-3) show the proposed approaches reduce the number of delay spikes by more than 50%.

A Deadline-Aware Scheduling Scheme for Connected Car Services Using Mobile Networks with Quality Fluctuation

Traffic collision is an extremely serious issue in the world today. The World Health Organization (WHO) reported the number of road traffic deaths globally has plateaued at 1.25 million a year. In an attempt to decrease the occurrence of such traffic collisions, various driving systems for detecting pedestrians and vehicles have been proposed, but they are inadequate as they cannot detect vehicles and pedestrians in blind places such as sharp bends and blind intersections. Therefore, mobile networks such as long term evolution (LTE), LTE-Advanced, and 5G networks are attracting a great deal of attention as platforms for connected car services. Such platforms enable individual devices such as vehicles, drones, and sensors to exchange real-time information (e.g., location information) with each other. To guarantee effective connected car services, it is important to deliver a data block within a certain maximum tolerable delay (called a deadline in this work). The Third Generation Partnership Project (3GPP) stipulates that this deadline be 100 ms and that the arrival ratio within the deadline be 0.95. We investigated an intersection at which vehicle collisions often occur to evaluate a realistic environment and found that schedulers such as proportional fairness (PF) and payload-size and deadline-aware (PayDA) cannot satisfy the deadline and arrival ratio within the deadline, especially as network loads increase. They fail because they do not consider three key elements— radio quality, chunk size, and the deadline—when radio resources are allocated. In this paper, we propose a deadline-aware scheduling scheme that considers chunk size and the deadline in addition to radio quality and uses them to prioritize users in order to meet the deadline. The results of a simulation on ns-3 showed that the proposed method can achieve approximately four times the number of vehicles satisfying network requirements compared to PayDA. key words: IoT, resource control, 5G, LTE, connected car, V2X