Delay-based Congestion Control for Multipath TCP

Space-ground integrated networks (SGINs) have been recently regarded as a promising way to provide resilient, dependable as well as efficient data transmission in the high-speed railway (HSR) scenario. Applying multipath transmission control protocol (MPTCP) to SGIN can realize data transmission simultaneously via terrestrial and satellite networks. However, since the existing congestion control (CC) mechanisms of MPTCP fail to distinguish between adverse influences (such as packet loss and/or round-trip time increase) caused by congestion and those caused by handovers, it suffers severe performance degradation in the HSR scenario where handover frequently occurs. In this article, we first present the SGIN oriented HSR (SGIN-HSR) with MPTCP. Then leveraging cross-layer information (i.e., reference signal received power), we design a novel cross-layer aided MPTCP CC mechanism targeted at SGIN-HSR based on deep reinforcement learning, which is referred to as HSR-CC, to alleviate performance degradation problems induced by handover. The experimental results show that HSR-CC significantly enhances the goodput and outperforms state-of-the-art MPTCP CC algorithms in SGIN-HSR environments where handover frequently occurs.

Multipath Transmission Control Protocol (MPTCP) enables multi-homed devices to establish multiple simultaneous routes for data transmission. Congestion Control (CC) is a fundamental mechanism for implementing and designing MPTCP. The Internet of Things (IoT) networks generate a massive volume of heterogeneous traffic with high dimensional states and diverse QoS characteristics. The existing MPTCP CC algorithms are unable to perform efficiently under highly mobile and dynamic IoT environments. We propose a novel model-free SDN-based adaptive actor-critic deep reinforcement learning framework based on a fuzzy normalized neural network to address the issue of CC for MPTCP in the IoT networks. In the proposed method, an agent can learn efficiently and better approximate the state-action value function of the actor and the action function of the critic to adjust the sub-flows congestion windows size adaptively according to the dynamic condition of a network. Simulation results show that the proposed scheme outperforms the state of the art schemes in terms of the goodput under highly-dynamic IoT environments.

With the development of the Internet of Things (IoT) and 5G, there are ubiquitous smart devices and network functions providing emerging network services efficiently and optimally through building many network connections based on WiFi, LTE/5G, Ethernet, and etc. The Multipath TCP (MPTCP) protocol that enables these devices to establish multiple paths for simultaneous data transmission, has been a widely used extension of standard TCP in smart devices and network functions. On the other hand, more heavy and time-varying traffic loads are generated in an MPTCP network, so that an efficient congestion control mechanism that schedules the traffic between multiple subflows and avoids congestion is highly required. In this paper, we propose a decentralized learning approach, DeepCC, to adapt to the volatile environments and realize the efficient congestion control. The Multi-Agent Deep Reinforcement Learning (MADRL) is used to learn a policy of congestion control for each subflow according to the real-time network states. To deal with the problem of the fixed state space and slow convergence, we adopt two self-attention mechanisms to receive the states and train the policy, respectively. Due to the asynchronous design of DeepCC, the learning process will not introduce extra delay and overhead on the decision-making process. Experiment results show that DeepCC consistently outperforms the well-known heuristic method and DRL-based MPTCP congestion control method in terms of goodput and jitter. Besides, DeepCC with the attention mechanism reduces convergence time by about 50% and increase goodput by about 80% compared with the commonly used structures of neural networks.

The last three decades have seen much evolution in web and network protocols: amongst them, a transition from HTTP/1.1 to HTTP/2 and a shift from loss-based to delay-based TCP congestion control algorithms. This paper argues that these two trends come at odds with one another, ultimately hurting web performance. Using a controlled synthetic study, we show how delay-based congestion control protocols (e.g., BBR and CUBIC + Hybrid Slow Start) result in the underestimation of the available congestion window in mobile networks, and how that dramatically hampers the effectiveness of HTTP/2. To quantify the impact of such finding in the current web, we evolve the web performance toolbox in two ways. First, we develop Igor, a client-side TCP congestion control detection tool that can differentiate between loss-based and delay-based algorithms by focusing on their behavior during slow start. Second, we develop a Chromium patch which allows fine-grained control on the HTTP version to be used per domain. Using these new web performance tools, we analyze over 300 real websites and find that 67% of sites relying solely on delay-based congestion control algorithms have better performance with HTTP/1.1.

Multipath TCP, as proposed by the IETF working group mptcp, allows a single data stream to be split across multiple paths. This has obvious benefits for reliability, and it can also lead to more efficient use of networked resources. We describe the design of a multipath congestion control algorithm, we implement it in Linux, and we evaluate it for multihomed servers, data centers and mobile clients. We show that some 'obvious' solutions for multipath congestion control can be harmful, but that our algorithm improves throughput and fairness compared to single-path TCP. Our algorithmis a drop-in replacement for TCP, and we believe it is safe to deploy.

TCP was originally designed for single session, thus could not utilize resources of multiple paths. MPTCP (Multipath TCP) provides enhanced performance and reliability by using multiple paths simultaneously at the Transport layer. But, LIA based congestion control of basic MPTCP does not show good performance when characteristics of two path are asymmetrical in the aspects of bandwidth in high-speed networks. In this paper, we suggest new MPTCP control scheme by applying TCP CUBIC on the high-speed path and modified LIA on the other low-speed data path. Our proposed scheme can enhance performance of high-speed asymmetric path more than 4 times compared to the basic MPTCP protocol.

Multipath TCP (MPTCP) is a promising TCP extension that exploits different Internet paths between a pair of hosts to obtain high aggregate throughput, while it also brings the concern of energy consumption. There has been a lively interest in the design of energy-efficient MPTCP. The research community, however, lacks a comprehensive understanding of which components in an MPTCP congestion control algorithm play the fundamental role in energy efficiency, how various algorithms compare against each other from energy-consuming perspective, or whether there exist potentially better solutions for energy saving. This paper takes the first step towards answering these questions. Through realworld MPTCP Linux kernel experiments, we show that the energy consumption is related to three major aspects: average throughput, path delay and different network scenarios. In order to bridge congestion control to the three aspects, we analyze the existing algorithms and capture the essential parameters of multipath congestion control model related to energy efficiency. Then we propose a window increase factor to shift traffic to low-delay energy-efficient paths. To further extend this design, we apply an energy-aware compensative parameter to fit the general hierarchical Internet topology. Our evaluation indicates that the enhanced congestion control module can successfully improve MPTCP energy efficiency without affecting its transmission performance.

To alleviate the lower performance of Transmission Control Protocol (TCP) congestion control over complex network, especially the high latency and packet loss scenario, Google proposed the Bottleneck Bandwidth and Round-trip propagation time (BBR) congestion control algorithm. In contrast with other TCP congestion control algorithms, BBR adjusted transfer data by maximizing delivery rate and minimizing delay. However, some evaluation experiments have shown that the persistent queues formation and retransmissions in the bottleneck can lead to serious fairness issues between BBR flows with different round-trip times (RTTs). They pointed out that small RTT differences cause unfairness in the throughput of BBR flows and flows with longer RTT can obtain higher bandwidth when competing with the shorter RTT flows. In order to solve this fairness problem, an adaptive congestion window of BBR is proposed, which adjusts the congestion window gain of each BBR flow in network load. The proposed algorithms alleviate the RTT fairness issue by controlling the upper limit of congestion window according to the delivery rate and queue status. In the Network Simulator 3 (NS3) simulation experiment, it shows that the adaptive congestion window of BBR (BBR-ACW) congestion control algorithm improves the fairness by more than 50% and reduces the queuing delay by 54%, compared with that of the original BBR in different buffer sizes.

A network rate management protocol with TCP congestion control and fairness for all

Previous theory and experiments show that delay- based algorithm, e.g. TCP Vegas, outperforms other congestion control algorithms with loss-based events. Unfortunately, previous analytic models of TCP Vegas always omit the impact of traffic in reverse path. To address this problem, we develop a simple analytic model in the presence of traffic in reverse path. This model captures the key innovative mechanisms that TCP Vegas employs during congestion avoidance and studies the relations between equilibrium window size and round trip time (RTT). In order to do this, We split RTT to two parts: forward trip time and reverse trip time. Results show that the delay time in reverse path has the same importance as forward data path in determining equilibrium window size and throughput of delay- based algorithm in congestion control.

In the past few decades, many Transmission Control Protocol (TCP) congestion control algorithms have been investigated to meet the growing network demands and to enhance the performance of TCP in lossy or high-bandwidth-delay-product (high-BDP) networks. However, it is still challenging to implement a dynamic congestion control algorithm for wide range of diverse mobile users, network conditions and applications. This paper explores avenues for enhancement of TCP congestion control algorithm for next generation mobile networks by dynamically learning the available bandwidth and deriving the Congestion Control Factor N. N is used to Adaptive Increase/Adaptive Decrease (AIAD) the Congestion Window (CWND) dynamically instead of using the traditional approach that is Additive Increase/ Multiplicative Decrease (AIMD) paradigm. Once there is congestion, our proposed algorithm will not allow the CWND to decrease multiplicatively or steeply. After dropping to a certain level (lower than legacy), we try to take the CWND to previous state adaptively with the help of calculated bandwidth (based on learning). This in turn helps to efficiently control the CWND for better network utilization especially in case of lossy and high-BDP conditions. As soon as it reaches the original state, it remains stable for longer time as compared to legacy until packet loss or time out. We demonstrate the effectiveness of our algorithm with the help of live air experiments (performed in Samsung R&D India, Bangalore) and NS3 based simulation experiments. Through our experiments, we show that our algorithm outperforms the legacy congestion control algorithms (like CUBIC, RENO, TCPW) and the existing CLTCP algorithm in terms of goodput, intra algorithm fairness and inter algorithm fairness maintaining the scalability and friendliness.

Satellite systems might become an important supplement to 5G systems. In addition, compared with regular TCP, multipath TCP (MPTCP) provides a robust solution. However, due to the high latency and high loss of satellite networks, most of the existing multipath TCP congestion control schemes do not perform well. Therefore, we proposed a bandwidth and round-trip time detection-based congestion control (BWRD) for multipath TCP over highly lossy satellite networks. Simulations are running to evaluate the proposed BWRD algorithm. It is shown that compared to previous solutions, our strategy significantly improves throughput performance and guarantees the fairness among the various sub-flows.

It has been reported that TCP Reno underutilizes network bandwidth, especially in High-Speed Long-Distance (HSLD) networks, while it has been widely adopted the dominant transport protocol in current Internet. Moreover, the Additive Increase Multiplicative Decrease (AIMD) congestion control algorithm of the TCP Reno has equilibrium and dynamic problems that become more and more severe as the bandwidth-delay product increases. An ideal TCP congestion control algorithm that achieves a high utilization, a small queueing delay, a stable behavior and fairness in bandwidth allocation has been major objectives of networking research in recent years. In this paper, we propose a new congestion control protocol, called TCP KWIK, a modification to the TCP Reno for HSLD networks. TCP KWIK uses delay-based feedback information to address these problems in order to stabilize a network around a fair and efficient operating point. We argue that the delay-based approach is better than the end-to-end congestion control as the networks capacity increases. Their advantage is small at low speed but decisive at high speed. The simulation results demonstrate the effectiveness of our proposed algorithm.KeywordsPacket LossCongestion ControlLoss ProbabilityRound Trip TimeCongestion WindowThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Nearly 50% of the energy overhead in today's datacenters comes from host-to-host data transfers, which largely depend on the transport layer performance. Multipath TCP (MPTCP) has recently been suggested as a promising transport protocol to improve datacenter network throughput, yet it also increases the host CPU power consumption. It remains unclear whether datacenters can indeed benefit from using MPTCP from the perspective of energy efficiency. By analyzing the performance of MPTCP, we show that (1) despite consuming higher host CPU power than TCP, MPTCP can largely reduce the long flow completion time and thus save the aggregated energy; (2) link-sharing subflows in MPTCP not only has negative impact on both throughput-sensitive long flows and latency-sensitive short flows, but also noticeably increases the host CPU power, especially for short flows. We present MPTCP-D, an energy-efficient variant of multipath TCP for datacenters. MPTCP-D incorporates a novel congestion control algorithm that can provide energy efficiency by minimizing the flow completion time, and an extra subflow elimination mechanism that can preclude link-sharing subflows from increasing the host CPU power. We implement MPTCP-D in the Linux kernel and study its performance by experiments on Amazon EC2. Our results show that, without degrading the performance of the long flow throughput and short flow completion time, MPTCP-D reduces the long flow energy consumption by up to 72% compared to DCTCP for data transfers, and reduces the short flow power consumption by up to 46% compared to MPTCP with linksharing subflows.

TCP congestion control has been designed to ensure Internet stability along with fair and efficient allocation of the network bandwidth. During the last decade, many congestion control algorithms have been proposed to improve the classic Tahoe/Reno TCP Congestion control. This paper aims at evaluating and comparing three control algorithms, which are Westwood+, New Reno and Vegas TCP. Simulation scenarios are carefully designed in order to investigate goodput, fairness and friendliness provided by each of the algorithms. results show that Westwood+ TCP is friendly towards New Reno TCP and improves fairness in bandwidth allocation whereas Vegas TCP is fair but is not able to grab its bandwidth share when coexisting with Reno or in the presence of reverse traffic because of its RTT-based congestion detection mechanism. Finally results show that Westwood* remarkably improves utilization of wireless links that are affected by losses not due to congestion.