Multipath TCP Research Articles

Fair and Efficient Distributed Edge Learning With Hybrid Multipath TCP

The bottleneck of distributed edge learning (DEL) over wireless has shifted from computing to communication, primarily the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">aggregation-averaging</i> (Agg-Avg) process of DEL. The existing transmission control protocol (TCP)-based data networking schemes for DEL are application-agnostic and fail to deliver adjustments according to application layer requirements. As a result, they introduce massive excess time and undesired issues such as unfairness and stragglers. Other prior mitigation solutions have significant limitations as they balance data flow rate from workers across paths but often incur imbalanced backlogs when the paths exhibit variance, causing stragglers. To facilitate a more productive DEL, we develop a hybrid multipath TCP (MPTCP) by combining model-based and deep reinforcement learning (DRL) based MPTCP for DEL that strives to realize quicker iteration of DEL and better fairness (by ameliorating stragglers). Hybrid MPTCP essentially integrates two radical TCP developments: i) successful existing model-based MPTCP control strategies and ii) advanced emerging DRL-based techniques, and introduce a novel hybrid MPTCP data transport for easing the communication of <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Agg-Avg</i> process. Extensive emulation results demonstrate that the proposed hybrid MPTCP can overcome excess time consumption and ameliorate the application layer unfairness of DEL effectively without injecting additional inconstancy and stragglers.

URLM: Utilising reinforcement learning to schedule subflows in MPTCP

SummaryParallel transmission in multiple access networks using MultiPath TCP (MPTCP) greatly enhances the throughput. However, critical packet disorder is commonly observed due to traffic fluctuation and path diversity. Although several predictive scheduling algorithms have been proposed to solve this problem, they cannot accommodate prediction accuracy and real‐time adaptation simultaneously in a dynamic network environment. The time overhead in modifying scheduling parameters to adapt to network changes leads to performance degradation in throughput and packet disorder. In this study, we propose a scheduling algorithm called Utilising Reinforcement Learning to Schedule Subflows in MPTCP (URLM). We apply reinforcement learning to select an optimal scheduling parameter in real time, which brings significant time benefits for modifying the parameters. The simulation comparison experiments show that URLM reduces the average number of out‐of‐order packets and the time overhead in adapting to network changes while improving global throughput.

Solving the multicommodity flow problem using an evolutionary routing algorithm in a computer network environment.

The continued increase in Internet traffic requires that routing algorithms make the best use of all available network resources. Most of the current deployed networks are not doing so due to their use of single path routing algorithms. In this work we propose the use of a multipath capable routing algorithm using Evolutionary Algorithms (EAs) that take into account all the traffic going over the network and the link capacities by leveraging the information available at the Software Defined Network (SDN) controller. The designed routing algorithm uses Per-Packet multipath routing to make the best use of the network's resources. Per-Packet multipath is known to have adverse affects when used with TCP, so we propose modifications to the Multipath TCP (MPTCP) protocol to overcome this. Network simulations are performed on a real world network model with 41 nodes and 60 bidirectional links. Results for the EA routing solution with the modified MPTCP protocol show a 29% increase in the total network Goodput, and a more than 50% average reduction in a flow's end-to-end delay, when compared to OSPF and standard TCP under the same network topology and flow request conditions.

Multipath TCP Meets Reinforcement Learning: A Novel Energy-Efficient Scheduling Approach in Heterogeneous Wireless Networks

Multipath TCP (MPTCP) has been standardized by the IETF as an extension of conventional TCP and it allows the system to utilize multiple paths simultaneously, which can aggregate bandwidth to improve network throughput. However, MPTCP needs to open multiple interfaces at the same time, which makes MPTCP consume more energy to maintain multiple interface connections. Thus, how to manage subflows with the MPTCP's scheduling system to determine which paths should be used for data transmission is of critical importance to reduce energy consumption and ensure network throughput. Due to the path heterogeneity and random packet losses in wireless networks, existing scheduling systems, and selecting paths based on the path's delay or energy cost, may suffer from performance degradation. In this article, we propose a reinforcement learning-based multipath scheduler called MPTCP-RL to determine the optimal path set for different flows. MPTCP-RL adopts deep reinforcement learning as well as MPTCP transmission model to manage path usage among multiple connections to make sure that the sender can adaptively select the optimal path set for a certain application according to the current network environment. MPTCP-RL is an asynchronous reinforcement learning framework, which separates the processes of offline training and online decision to ensure that the learning process will not introduce extra delay and overhead on the decision making process in MPTCP path management. The extensive experimental results show that MPTCP-RL can improve the aggregate throughput and reduce energy consumption significantly compared to the state-of-the-art mechanisms in a variety of network scenarios.

Multipath TCP, and New Packet Scheduling Method

Multipath TCP, and New Packet Scheduling Method

An implementation and evaluation of MPTCP-based IoT router.

The evolving Internet of Things (IoT) promisingly improves the quality of life and transforms many industries. However, the IoT application challenges the wireless networks since the resource-constrained IoT devices typically need to send data to the cloud or edge server. Therefore, it is necessary to introduce an intermediate device between IoT devices and the servers, for example, to reduce the cost of direct communication between them. In another case, the device may move and collect the data from IoT devices before transmitting it to the server. The intermediate device should be designed to have resilient Internet connections and sufficient bandwidth in such a context. This work implements and evaluates a Multipath TCP (MPTCP) IoT router, which uses multiple radios to connect a server to address the demanding design. The router leverages MPTCP, an extension of TCP for simultaneous transmission over several paths on top of Wi-Fi interfaces. MPTCP has also supported several working modes for throughput and (or) resilience enhancements. First, we implement the MPTCP kernels, which can run on the popular IoT devices Raspberry Pi 3B+ and 4. Second, we extensively evaluate the performance of IoT routers in a static and mobility scenario. The static scenario's evaluation results show that the MPTCP-based router can achieve seamless handover and bandwidth aggregation. In the mobility scenario, the MPTCP router with one backup path performs better than the single-path TCP. Besides, the MPTCP routers are more energy-efficient than TCP on the same hardware.

Achieving Flexible and Lightweight Multipath Congestion Control Through Online Learning

The upgrade of network devices to be equipped with multiple network interfaces makes it possible to improve network throughput performance through multipath transmission protocols, especially multipath TCP (MPTCP). However, so far the mostly used MPTCP protocols have a common limitation, namely the rigid and conservative method. They have been designed with little consideration of the fact that real networks are dynamic and the network status changes frequently, thus leading to the poor performance of current MPTCP in many realistic scenarios. In this paper, we propose a lightweight multipath congestion control algorithm based on online learning, named MP-OL. MP-OL models congestion control as a multi-armed bandit problem, and adjusts the sending rate of each subflow flexibly and adaptively through online learning. Therefore, MP-OL possesses the capability of suiting various network scenarios, and can achieve fairness and high performance in dynamic network environment. It can also flexibly switch between online learning and traditional method, which reduces the computational complexity while ensuring the learning efficiency, thus making MP-OL easy to deploy and use. As the experimental results demonstrated, compared with the leading MPTCP variants, MP-OL achieves significant improvements in fairness and link utilization, and shows better resilience to non-congestion loss and better adaptability to unstable network conditions. In real networks, MP-OL also obtains better throughput performance.

Learning to Harness Bandwidth With Multipath Congestion Control and Scheduling

Multipath TCP (MPTCP) has emerged as a facilitator for harnessing and pooling available bandwidth in wireless/wireline communication networks and in data centers. Existing implementations of MPTCP such as, Linked Increase Algorithm (LIA), Opportunistic LIA (OLIA) and BAlanced LInked Adaptation (BALIA) include separate algorithms for congestion control and packet scheduling, with pre-selected control parameters. We propose a Deep Q-Learning (DQL) based framework for joint congestion control and packet scheduling for MPTCP. At the heart of the solution is an intelligent agent for interface, learning and actuation, which learns from experience optimal congestion control and scheduling mechanism using DQL techniques with policy gradients. We provide a rigorous stability analysis of system dynamics which provides important practical design insights. In addition, the proposed DQL-MPTCP algorithm utilizes the ‘recurrent neural network’ and integrates it with ‘long short-term memory’ for continuously i) learning dynamic behavior of subflows (paths) and ii) responding promptly to their behavior using prioritized experience replay. With extensive emulations, we show that the proposed DQL-based MPTCP algorithm outperforms MPTCP LIA, OLIA and BALIA algorithms. Moreover, the DQL-MPTCP algorithm is robust to time-varying network characteristics, and provides dynamic exploration and exploitation of paths.

Research on the network handoff strategy based on the best access point name decision.

To improve the network switching performance and efficiency of mobile phone terminals and establish an efficient mobile communication network connection, this paper constructs the SDN+MPTCP+CP (Software Defined Network, Multi-Path TCP, and Mobile Terminal) mobile communication network model and designs a network switching algorithm with a preselected available access point name (APN) based on the potential game method. The constructed network model integrates a 5G mobile communication network, satellite communication, the SDN network, and the MPTCP multi-way communication technology. APN access point is preselected by using Kalman filtering theory, and dual problems are resolved with the Lagrange function. To determine the MPTCP sub-flow transmission path, the differential derivative calculation is introduced. The performance of the network switching strategy is evaluated based on the Jacobian matrix. Then, the game coefficient is designed, and the game function is calculated. A potential game balance point is found, and an updating strategy is formulated to determine the best APN access point. The simulation network model is constructed, and the parameters of the performance evaluation are defined to find the performance comprehensively. The experimental results demonstrated the extreme reliability, stability, and compatibility of the proposed algorithm.

Coupled CUBIC Congestion Control for MPTCP in Broadband Networks

Recently, multipath transmission control protocol (MPTCP) was standardized so that data can be transmitted through multiple paths to utilize all available path bandwidths. However, when high-speed long-distance networks are included in MPTCP paths, the traffic transmission performance of MPTCP is severely deteriorated, especially in case the multiple paths’ characteristics are heavily asymmetric. In order to alleviate this problem, we propose a “Coupled CUBIC congestion control” that adopts TCP CUBIC on a large bandwidth-delay product (BDP) path in a linked increase manner for maintaining fairness with an ordinary TCP traversing the same bottleneck path. To verify the performance excellence of the proposed algorithm, we implemented the Coupled CUBIC Congestion Control into Linux kernels by modifying the legacy MPTCP linked-increases algorithm (LIA) congestion control source code. We constructed asymmetric heterogeneous network testbeds mixed with large and small BDP paths and compared the performances of LIA and Coupled CUBIC by experiments. Experimental results show that the proposed Coupled CUBIC utilizes almost over 80% of the bandwidth resource in the high BDP path, while the LIA utilizes only less than 20% of the bandwidth for the same path. It was confirmed that the resource utilization and traffic transmission performance have been greatly improved by using the proposed Coupled CUBIC in high-speed multipath networks, as well as maintaining MPTCP fairness with competing single-path CUBIC or Reno TCP flows.

Toward the Implementation of MPTCP Over mmWave 5G and Beyond: Analysis, Challenges, and Solutions

5G and beyond 5G networks are going to be an inseparable part of human lives in the future. They will dominate any aspects of everyday activities from smart homes, remote surgery, and smart cities to autonomous driving, which demand high throughput through low latency end-to-end communication. Consequently, employing higher frequencies in millimeter-wave for the coming cellular networks will be inevitable. On the one hand, millimeter-wave can provide massive data rates; on the other hand, it suffers from some shortcomings such as blockage and misalignment that can occur because of the susceptible characteristic of the high frequencies. These flaws can mislead TCP in adjusting its sending rate efficiently and reduce the connection quality that a user discerns. Deploying Multipath TCP (MPTCP) is one of the schemes that can relieve the aforementioned issues and assist in utilizing the new generation mobile networks’ high potential by leveraging its features in exploiting diverse paths in networks such as NR or LTE. A well-designed MPTCP can select the best path among the available ones, so it can enhance the perceived user experience. This paper fills the gap of having a comprehensive overview of MPTCP and its deployment over 5G and beyond 5G networks. It analyses millimeter-wave-based cellular networks, MPTCP procedures and parameters, state-of-the-art MPTCP congestion control mechanisms and schedulers, and the probable solutions that may enhance the protocol’s functionality in cellular communication.

Multipath Congestion Control: Measurement, Analysis, and Optimization From the Energy Perspective

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.

Internet of Drones: Improving Multipath TCP over WiFi with Federated Multi-Armed Bandits for Limitless Connectivity

We consider multipath TCP (MPTCP) flows over the data networking dynamics of IEEE 802.11ay for drone surveillance of areas using high-definition video streaming. Mobility-induced handoffs are critical in IEEE 802.11ay (because of the smaller coverage of mmWaves), which adversely affects the performance of such data streaming flows. As a result of the enhanced 802.11ay network events and features (triggered by beamforming, channel bonding, MIMO, mobility-induced handoffs, channel sharing, retransmissions, etc.), the time taken for packets to travel end-to-end in 802.11ay are inherently time-varying. Several fundamental assumptions inherent in stochastic TCP models, including Poisson arrivals of packets, Gaussian process, and parameter certainty, are challenged by the improved data traffic dynamics over IEEE 802.11ay networks. The MPTCP model’s state estimation differs largely from the actual network values. We develop a new data-driven stochastic framework to address current deficiencies of MPTCP models and design a foundational architecture for intelligent multipath scheduling (at the transport layer) considering lower layer (hybrid) beamforming. At the heart of our cross-layer architecture is an intelligent learning agent for actuating and interfacing, which learns from experience optimal packet cloning, scheduling, aggregation, and beamforming using successful features of multi-armed bandits and federated learning. We demonstrate that the proposed framework can estimate and optimize jointly (explore–exploit) and is more practicable for designing the next generation of low-delay and robust MPTCP models.

Low Latency and High Data Rate (LLHD) Scheduler: A Multipath TCP Scheduler for Dynamic and Heterogeneous Networks.

The scheduler is a crucial component of the multipath transmission control protocol (MPTCP) that dictates the path that a data packet takes. Schedulers are in charge of delivering data packets in the right order to prevent delays caused by head-of-line blocking. The modern Internet is a complicated network whose characteristics change in real-time. MPTCP schedulers are supposed to understand the real-time properties of the underlying network, such as latency, path loss, and capacity, in order to make appropriate scheduling decisions. However, the present scheduler does not take into account all of these characteristics together, resulting in lower performance. We present the low latency and high data rate (LLHD) scheduler, which successfully makes scheduling decisions based on real-time information on latency, path loss, and capacity, and achieves around 25% higher throughput and 45% lower data transmission delay than Linux's default MPTCP scheduler.

${l}\,^2$-MPTCP: A Learning-Driven Latency-Aware Multipath Transport Scheme for Industrial Internet Applications

With various industrial wireless networks greeting booming development, modern industrial devices configured with several network interfaces increasingly become the norm. Such multihomed industrial devices can increase application throughput by making use of multiple network paths, enabled by the multipath transmission control protocol (MTCP) (MPTCP). However, MPTCP might be challenged in the heterogeneous industrial networks because concurrent transmitting industrial application data over asymmetric network paths with different delays is almost bound to the receive buffer blocking problem, which is caused by out-of-order packet arrival and is harmful to the performance of the multipath transmission. The existing MPTCP solutions generally use static mathematical models to evaluate path quality and prohibit transmission on paths with poor quality, which are unable to perform efficiently under highly dynamic and complex network environments. Therefore, in this article, we propose a learning-driven latency-aware MPTCP variant, called <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">${l}\,^2$</tex-math></inline-formula> -MPTCP, which seeks to possibly mitigate the out-of-order packet arrival and receive buffer blocking problems associated with the network heterogeneity in the industrial Internet. <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">${l}\,^2$</tex-math></inline-formula> -MPTCP accurately computes each MPTCP path’s forward delay and assigns application data to multiple paths according to their calculated forward delay differences by using a novel multiexpert learning-enabled forward delay estimator. <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">${l}\,^2$</tex-math></inline-formula> -MPTCP dynamically manages path usage and chooses the optimal path collection for bandwidth aggregation and multipath transmission by using a promising reinforcement learning-empowered multipath manager. Experimental results demonstrate that <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex-math notation="LaTeX">${l}\,^2$</tex-math></inline-formula> -MPTCP outperforms the current MPTCP solutions in terms of multipathing service quality.

Adaptive QoS-aware multipath congestion control for live streaming

Nowadays, the popularization of live streaming bring challenges in providing customized Quality of Service (QoS) and satisfying diverse Quality of Experience (QoE) requirements. Multipath TCP (MPTCP), designed for bandwidth and reliability in aggregating transmission, has the potential to improve live streaming performance with multiple simultaneously transmitting paths. However, the existing multipath congestion control algorithms (CCAs) of MPTCP fail to adapt to diverse network environments and different QoS requirements. We study the problem with extensive experiments to observe the limitations of current multipath CCAs. To tackle these problems and support stable and customized live streaming, we propose ACCeSS, an adaptive QoS-aware multipath congestion control framework. ACCeSS is able to promptly adapt to network changes and QoS requirements with a novel control policy optimization phase. In order to adjust and stimulate improvement on the preferred performance metric, ACCeSS exploits Random Forest Regressing (RFR) method to perform QoS-specific utility function optimization. Besides, ACCeSS is implemented and deployed in a multipath live streaming system. We compare it with other multipath CCAs in the Linux kernel and evaluate their performance in both emulated and real-world networks. It is revealed that ACCeSS outperforms classic multipath CCAs and the state-of-the-art learning-based multipath CCA, with better environment adaptive capability of QoS and higher QoE for live streaming.

Technical Comparison between MPTCP and TCP in Heterogeneous Networks

Multipath TCP (MPTCP) is a project that has proposed by IETF group to allow packet streams to be separated across multiple sub-flow paths. MPTCP brings a great benefit in throughput, reliability, and fairness. However, MPTCP can be implemented in a little Linux distribution where it requires compiling and installing to become available for an actual and simulation scenario. In the paper, the performance analysis and evaluation results in both TCP and MPTCP are achieved through PC computer connected to Wi-Fi access point and 3G in heterogeneous networks. The simulation results prove that MPTCP performs better than single-path TCP for heterogeneous networks. All experiments over MPTCP and TCP leads to same results in terms of the throughput in heterogeneous networks. In this paper, four congestion control methods are implemented in the kernel and compared with each other understand the behaviors of those methods. The result of the experiment proves that Linked Increased Algorithm (LIA) has the advantage over the others in light data traffic while Balanced LIA (BLIA) has the advantage the others when there is heavy data traffic

An Analysis of MPTCP Congestion Control

Many devices contain more than one network interface. There is scope for multi-path transfer to utilise these network interfaces simultaneously. Multi-path TCP (MPTCP) is designed to provide improved resilience and resource utilisation through multi-path transfer. One of the key goals of MPTCP is to preserve fair resource sharing with regular TCP at network bottlenecks. Although the coupled congestion control algorithms can achieve this goal by coupling subflow congestion windows, the algorithms always assume that the subflow paths will share a bottleneck. As a consequence, MPTCP is unable to maximise throughput over all available paths at a non-shared bottleneck. We present a survey about MPTCP and its coupled congestion control algorithms. We then show that MPTCP coupled congestion control algorithms perform poorly when paths are disjoint and/or do not have similar delay and/or bandwidth characteristics.

Modified Multipath Transmission Control Protocol Using Logarithmic Increase Under Long-Term Evolution and Wi-Fi Networks

Multipath transmission control protocol (MPTCP) has been proposed for devices with multiple network interfaces. This idea seems to be impossible for user implementation at the beginning of its proposal. However, the emergence of smartphones with multiple network access allows this idea to be implemented. This paper provides both theoretical and empirical throughput analysis of MPTCP under long term evolution (LTE) and Wi-Fi networks. First, a new modified congestion control is presented as an enhancement of standard TCP Reno using logarithmic function. Second, the existing MPTCP approaches like linked increase algorithm (LIA) and balanced LIA (BALIA) are discussed briefly. Finally, the experimental results under long term evolution (LTE) and Wi-Fi under software defined wireless networking (SDWN) are given along with the analysis. The experimental result shows that the proposed congestion control achieves higher average congestion window size. The BALIA MPTCP is shown to have higher throughput than that of the LIA MPTCP.

Performance Evaluation of MPTCP on Simultaneous Use of 5G and 4G Networks.

The 5G cellular network comes with a promise to provide a very high data rate at low latency, which is becoming critical for advancing technologies. Mobile operators are currently deploying the 5G cellular network worldwide. However, because of limited coverage and high susceptibility of the 5G network to obstacles, handoffs from 5G to 4G and vice versa frequently occur, especially when the user equipment (UE) is moving. These handoffs often cause significant delays in data transmission due to packet losses and retransmissions. A promising solution can be to use both 4G and 5G networks simultaneously, which can solve this problem and yield a better throughput. Multipath transmission control protocol (TCP) is an effective solution for this problem, but it requires significant performance evaluation before practical deployment. In this study, we implement an MPTCP testbed based on NS3-DCE that enables to test the performance of MPTCP schedulers and congestion control algorithms (CCAs) in both 3GPP and non-3GPP networks. Through extensive simulation experiments in a scenario where a UE simultaneously utilizes both 4G and 5G networks, we found that blocking estimation (BLEST) scheduler implemented with balanced linked adaptation (BALIA) CCA can produce the highest throughput and lowest delay. Finally, we showed how received signal to interference and noise ratio (SINR), congestion window, throughput, and packet losses are interconnected.