Additive Increase Multiplicative Decrease Research Articles

DECC: Achieving Low Latency in Data Center Networks With Deep Reinforcement Learning

Data Center Networks (DCNs) suffer from synchronized bursts for segmentation and aggregation modes, leading to buffer overflows at switches and increasing network delay. To overcome this problem, some congestion control algorithms like DCTCP use Explicit Congestion Notification (ECN) to notify in-network congestion and reduce switch buffer occupancy. However, the traditional Additive Increase Multiplicative Decrease (AIMD) method causes high fluctuation of round-trip time (RTT) in DCNs. Some intelligent congestion control algorithms designed for Internet can achieve great flexibility, but are not applicable in DCNs for a lack of accurate congestion feedback. In this paper, we analyze the deficiencies of utilizing RTT as congestion signals and the applicability of learning algorithms in DCNs. Then, we propose DECC, a smart TCP congestion control algorithm for DCNs, which combines Deep Reinforcement Learning (DRL) with ECN to achieve high bandwidth utilization as well as low queuing delay. DECC fully utilizes precise in-network feedback and formulates several QoS requirements to a multi-objective function. Meanwhile, it decouples cwnd adjustment with DRL decision making to gradually learn the optimal congestion control policy in real-time. We evaluate the performance of DECC in various scenarios. Simulation results show that DECC can reduce the queue length at bottleneck switches by more than 50% compared to DCTCP, while maintaining high bandwidth utilization and reducing Flow Completion Time (FCTs) under burst traffic.

Formation of Innovative Algorithm for High Speed Protocol using Network Simulator

The effort delivered in this manuscript is grounded on NS-2 (The Network Simulator 2) to implementing the congestion control process of classic TCP (Transmission Control Protocol), with new congestion control mechanism. In this paper, a novel congestion control algorithm is offered, which contains of slow-start and congestion avoidance mechanisms. The proposed slow-start algorithm assumes a duplicating and an interpolating approach to the congestion window (cwnd) for each increment instead of the exponential increment used by other TCP source variants such as Reno, Vega, Tahoe, Newreno, Fack, and Sack. Furthermore, the enhanced congestion avoidance algorithm is built by using an improved Additive Increase Multiplicative Decrease (AIMD) algorithm with multi TCP flow facility, to provide an enhanced congestion control algorithm with some valuable properties to improving TCP routine for high speed protocols.

WOA-Based Robust Congestion Control Scheme with Two Kinds of Propagation Latencies and External Disturbance in Software-Defined Wireless Networks

This paper proposes a novel WOA-based robust control scheme with two kinds of propagation latencies and external disturbance implemented in Software-Defined Wireless Networks (SDWNs) to maximize overall throughput and enhance the stability of the global network. Firstly, an adjustment model developed using the Additive-Increase Multiplicative-Decrease (AIMD) adjustment scheme with propagation latency in device-to-device paths and a closed-loop congestion control model with propagation latency in device–controller pairs are proposed, and the effect of channel competition from neighboring forwarding devices is analyzed. Subsequently, a robust congestion control model with two kinds of propagation latencies and external disturbance is established. Then, a new WOA-based scheduling strategy that considers each individual whale as a specific scheduling plan to allocate appropriate sending rates at the source side is presented to maximize the global network throughput. Afterward, the sufficient conditions are derived using Lyapunov–Krasovskii functionals and formulated using Linear Matrix Inequalities (LMIs). Finally, a numerical simulation is conducted to verify the effectiveness of this proposed scheme.

AIMD Rule-Based Duty Cycle Scheduling in Wireless Sensor Networks Using Quartile-Directed Adaptive Genetic Algorithm

Wireless sensor networks (WSNs) are attractive, in large part because they do not require wired infrastructure. However, this very characteristic renders them energy-constrained. Duty cycle scheduling (DCS) is thought to contribute to the energy efficiency of sensing. We develop a novel paradigm for modeling WSNs and a DCS strategy based on the additive increase/multiplicative decrease (AIMD) rule. The scheduling problem is framed as an optimization problem with the optimization objectives of lowering energy consumption and enhancing detection ability. The classic GA converges slowly in late evolution phases for using fixed crossover and mutation probability. An adaptable genetic algorithm (AGA) eliminates the above concern by allowing modifying crossover and mutation probability as per fitness values, but, unfortunately, it often fails to select the best individuals. We suggest a quartile-directed adaptive genetic algorithm (Q-AGA) in which individuals with fitness above the first quantile of the last generation were selected. The interquartile range (IQR) is utilized to weight crossover and mutation probability. If IQR is larger, the fitness distribution is more concentrated, and then, we can diminish the crossover and mutation probability, allowing favorable traits to be retained to the greatest extent feasible. Simulation results show that Q-AGA has a faster convergence speed and a stronger global search ability. Furthermore, we change the hypothesis of the system model by considering that the two objectives can be weighted, thereby broadening the application scenarios.

The Value of Cooperation

The well-known Additive Increase-Multiplicative Decrease (AIMD) abstraction for network congestion control was first published by Dah-Ming Chiu and Raj Jain in their seminal work [4] in 1989 and soon played a prominent part in TCP algorithm design for the Internet. The ingenuity of AIMD lies in the abstraction of Internet congestion control, and ever since its inception has also been a staple part of teaching curriculum for performance evaluation and computer networking courses at universities worldwide. In this paper, we describe teaching examples for university students to appreciate the AIMD abstraction from the theoretical aspects such as convex optimization and Perron-Frobenius theory to the data science aspect. The essence of cooperation encompassed by AIMD reverberates even in teaching networks formed by students and educators, giving rise to online classroom flipping teaching tools and data analytics to close the gap between teachers and students.

Blockchain Based Privacy Preserving Authentication and Malicious Node Detection in Internet of Underwater Things (IoUT) Networks

Internet of Underwater Things (IoUT) Networks are used to sense different aquatic parameters like temperature, pressure, pollution, etc. They are also used to forecast the ocean’s weather to collect information about natural disasters. However, they are easily compromised by attackers due to deployment in unattended environments. To overcome these issues, security is required in IoUT networks to avoid unauthorized access and ensure network credibility. This work proposes an authentication and a malicious node detection mechanism to restrict the unauthorized external nodes from accessing the network and the internal nodes from acting maliciously, respectively. Moreover, blockchain stores the hashes of sensor nodes’ credentials during the registration process to make the system secure and traceable. Meanwhile, a weighted trust evaluation mechanism is implemented for data aggregation and detection of malicious nodes. Moreover, an additive increase multiplicative decrease algorithm puts malicious nodes in an intensive observation queue to verify the data of malicious nodes before aggregating. Moreover, weights are assigned to sensor nodes based on their behaviour. If the weight of a sensor node becomes zero, it is revoked by the blockchain. The simulation results show the efficiency of our proposed malicious node detection mechanism in terms of energy consumption and propagation delay.

Distributed Nonlinear AIMD Algorithms for Electric Bus Charging Plants

Recently, the introduction of electric vehicles has given rise to a new paradigm in the transportation field, spurring the public transport service in the direction of using completely electric bus fleets. In this context, one of the main challenges is that of guaranteeing an optimal scheduling of the charging process, while reducing the power supply requested from the main grid, and improving the efficiency of the resource allocation. Therefore, in this paper, a power allocation strategy is proposed in order to optimize the charging of electric bus fleets, while fulfilling the limitation imposed on the maximum available power, as well as ensuring limited charging times. Specifically, relying on real bus charging scenarios, a charging optimization algorithm based on a Nonlinear Additive Increase Multiplicative Decrease (NAIMD) strategy is proposed and discussed. This approach is designed on the basis of real charging power curves related to the batteries of the considered vehicles. Moreover, the adopted NAIMD algorithm allows us to minimize the sum of charging times in the presence of saturation constraints in a distributed way and with a small amount of aggregated data sent over the communication network. Finally, an extensive simulation campaign is illustrated, showing the effectiveness of the proposed approach both in allocating the power resources and in sizing the maximum power capacity of charging plants in progress.

Leave No Cash on the Table: An Optimal Approach for Controlling Wireless TCP AIMD

The class of additive-increase/multiplicative-decrease (AIMD) algorithms constitutes a key mechanism for congestion control in modern communication networks, such as the current Internet. However, when wireless links are present, the algorithmic behavior of an AIMD algorithm is distorted. Spurious backoffs may be triggered due to packet reordering and non-congestive loss. Prior solutions are developed based on empirical observations without a complete examination of the available design options, thereby overlooking more efficient algorithmic designs (i.e. cash on the table). In this paper, we provide a systematic approach for attaining an optimal enhancement of wireless TCP AIMD. We formulate the design of wireless TCP AIMD as an optimal control problem with the design space of TCP AIMD as the feasible set. With a complete, formal characterization of the design space, we derive a compensation scheme that gives rise to a family of the optimal AIMD algorithms, which can fully filter out the effects due to packet reordering and non-congestive loss. We further identify some properties that help explain the impact of AIMD-based TCP enhancements systematically on the network. We have devised a simple module that realizes the compensation scheme. The module demonstrates a significant performance improvement over existing wireless TCP enhancements through packet level simulation.

A Full-Process Optimization-Based Background Subtraction for Moving Object Detection on General-Purpose Embedded Devices

Real-time computer vision tasks are emerging in consumer electronics with lightweight computing performance, which are an exquisite design art to balance the computational efficiency and accuracy. In this paper, we present the embedded background subtraction (EBGS) - an optimization algorithm for the entire process to increase computational efficiency and detection accuracy simultaneously. EBGS exploits a simple and efficient Additive Increase Multiplicative Decrease (AIMD) filter to improve the foreground detection accuracy without spending too much time. Moreover, the design combination between the contracted codebook background subtraction (BGS) model and a random model update is proposed to reduce the time consumption. Experiments demonstrate that EBGS can decrease the computing overhead for the three parts of BGS process simultaneously and achieve real-time performance and satisfactory detection accuracy under challenging environments.

Congestion control in high‐speed networks using the probabilistic estimation approach

SummaryNowadays, the bulk of Internet traffic uses Transmission Control Protocol (TCP) for reliable transmission. But the standard TCP's performance is very poor in high‐speed networks (HSNs), and hence, the core gigabytes links are usually underutilization. This problem has roots in conservative nature of TCP, especially in its Additive Increase Multiplicative Decrease (AIMD) phase. In other words, since TCP cannot figure out precisely the congestion status of the network, it follows a conservative strategy to keep the network from overwhelming. We believe that precisely congestion estimation in the network can solve this problem by avoiding unnecessary conservation. To this end, this paper proposes an algorithm, which considers packet loss and delay information jointly and employs a probabilistic approach to accurately estimation of congestion status in the network. To examine the proposed scheme performance, extensive simulations have been performed in the NS‐2 environment. Simulation results reveal that the proposed algorithm has better performance than existing algorithms in terms of bottleneck utilization, stability, throughput, and fairness.

Decentralized Charging of Plug-In Electric Vehicles and Impact on Transmission System Dynamics

This article focuses on the impact of the charge of Plug-in Electric Vehicles (PEVs) on the dynamic response of power systems and proposes an efficient solution to control electric vehicle chargers, by dynamically allocating the available power in an optimized way. The proposed approach is based on an Additive-Increase-Multiplicative-Decrease (AIMD) stochastic decentralized control strategy to efficiently and seamlessly manage the charge of a high number of PEVs with little communication efforts. A modified version of the New England network is utilized to validate the proposed control through a variety of scenarios and control setups.

Shared bottleneck detection based on trend line regression for multipath transmission

SummaryCurrent proposed congestion control algorithms for multipath transmission couple all subflows together to maintain bottleneck fairness by assuming that the subflows traverse a common bottleneck. The coupled form congestion control algorithms restrict the multipath session to gain better throughput in non‐shared bottleneck situation. The throughput of a multipath session following independent rate control is larger than the throughput achieved by coupled form congestion control algorithm. The key point is to find a method to detect whether flows share a common bottleneck. If the subflows of a multipath session are detected out traversing a common bottleneck, the sender could perform coupled congestion control algorithm to guarantee bottleneck fairness and follow independent congestion control algorithm otherwise to improve throughput. Deduced from fluid model of additive increase multiplicative decrease (AIMD) rule, the delay samples can be linearly fitted when the bottleneck link falls into congestion. A delay trend line regression (TR) method is proposed for shared bottleneck detection. The delay slopes in close proximity indicate that the flows traverse the same bottleneck in great probability. Extensive simulations are performed to validate the effectiveness of TR algorithm in shared bottleneck detection. It is further implemented in multipath QUIC Golang codebase to work in collaboration with linked increases algorithm (LIA). Results show a multipath QUIC session with two subflows can gain higher throughput in non‐shared bottleneck situation and is less aggressive in shared bottleneck cases with the implementation of shared bottleneck detection algorithm.

Decentralized Additive Increase and Multiplicative Decrease-Based Electric Vehicle Charging

Electric vehicle (EV) transition and low-cost renewable energy generation are putting power grid under a challenging transformation. Number of power electronics actuators connected to the grid is increasing, and the legacy control methods employed on the grid are not responsive to this growing demand. Thus, the grid integration of EVs and their charging management requires a system-wide solution that is scalable, autonomous, and stable. In this article, we investigate two very complex networks: Internet and power grid in the context of controlling mass-scale EV charging problem. We adapt the well-known additive increase-multiplicative decrease (AIMD) algorithm used in the Internet congestion control to EV charging in a distributed fashion. We develop an adaptation of the Internet's congestion control method for power grid considering the unique grid constraints using a decentralized concept. The advantage of the proposed method lies in its low-cost (memory-less) congestion detection mechanism based on only local voltage measurements. Results show that decentralized AIMD can successfully help flatten the peak loading caused by high EV penetration. To test the algorithm, a distribution grid model is designed based on IEEE 37-node test feeder with realistic load modeling. Finally, the results are presented in comparison with two other control architectures.

An Optimized Decentralized Power Sharing Strategy for Wind Farm De-Loading

Many centralized and distributed power sharing algorithms have been proposed in the literature for de-loading operations in wind farms with variable speed wind turbines. Typically, in these strategies, two-way communications are required between the control center and the single turbines, or among the turbines. This paper solves the same problem in a truly decentralized fashion, which only requires a greatly reduced amount of one-way communications, without exchanging information among the turbines, and shows that an optimal solution can be obtained to minimize utility functions of interest of single wind turbines. In particular, we consider utility functions that take into account mechanical fluctuations and rotor over-speeds during transient, while balancing the utilization of wind turbines at steady-state operations. This is achieved by adopting the so-called Additive Increase Multiplicative Decrease (AIMD) algorithms, which are frequently used in communication applications, for solving the power sharing problem in a decentralized fashion. Extensive simulations under different working conditions, on wind farms consisting of wind turbines of different mechanical characteristics, are provided to illustrate the potential and the efficiency of the proposed methodology.

Development of an efficient mechanism for rapid protocols using NS-2 simulator

The delivered effort in this manuscript is grounded on NS-2 (The Network Simulator 2) to implement the congestion control process of classic TCP (Transmission Control Protocol), with new congestion control mechanism. In this paper, a novel congestion control algorithm is offered, which contains of slow-start and congestion avoidance mechanisms. The proposed slow-start algorithm assumes a duplicating and an interpolating approach to the congestion window (cwnd) for each increment instead of the exponential increment used by other TCP source variants such as Reno, Vega, Tahoe, Newreno, Fack, and Sack. Furthermore, the enhanced congestion avoidance algorithm is built by using an improved Additive Increase Multiplicative Decrease (AIMD) algorithm with multi TCP flow facility, to provide an enhanced congestion control algorithm with some valuable properties to improve TCP routine for high speed protocols. The improvement strategy based on merging of slow start, congestion avoidance mechanism that are used in TCP congestion control, to create a new AIMD algorithm with a new relationship between the pair parameters a and b. This paper is also involved in the creation of rapid agent in NS-2 models designed to identify the modified TCP and to configure the NS-2 platform. A fast TCP also includes an innovative scheme to slow the rapid start to help TCP to start faster through the high speed networks and also to postpone the congestion state as much as possible.

A caching strategy based on dynamic popularity for named data networking

Named data networking (NDN) is a prominent architecture for the future internet. In NDN, routers have the capacity of in-network cache, which can completely improve network performance. However, the cache capacity in routers is limited and how to utilise the cache resources effectively is still a great challenge. To solve the problem, this study presents a dynamic popularity caching strategy based on additive increase multiplicative decrease for NDN (DPCA). DPCA takes content popularity and caching capacity into account and it utilises AIMD algorithm to adjust the popularity threshold dynamically. At the same time, it also proposes a evict algorithm which takes the historical information of content popularity, the trend of content request and the interval from the last request time into account. The simulation results show that the DPCA strategy can effectively improve cache hit ratio, decrease network throughput and reduce the average hit distance compared with other schemes.

A caching strategy based on dynamic popularity for named data networking

Named data networking (NDN) is a prominent architecture for the future internet. In NDN, routers have the capacity of in-network cache, which can completely improve network performance. However, the cache capacity in routers is limited and how to utilise the cache resources effectively is still a great challenge. To solve the problem, this study presents a dynamic popularity caching strategy based on additive increase multiplicative decrease for NDN (DPCA). DPCA takes content popularity and caching capacity into account and it utilises AIMD algorithm to adjust the popularity threshold dynamically. At the same time, it also proposes a evict algorithm which takes the historical information of content popularity, the trend of content request and the interval from the last request time into account. The simulation results show that the DPCA strategy can effectively improve cache hit ratio, decrease network throughput and reduce the average hit distance compared with other schemes.

An Internet-Inspired Proportional Fair EV Charging Control Method

Transportation systems are undergoing a major transition with the integration of electric vehicles (EVs). However, due to increase in battery energy and charger power ratings, potential adverse effects on the distribution grid is a crucial issue to be addressed. Large voltage drops at charging nodes will deteriorate the quality of power service and cause unfair utilization of grid capacity among EV users. Safe and efficient operation of the grid along with a fast, convenient, and fair charging strategy is an important research problem. In this paper, we adapt the additive increase multiplicative decrease (AIMD) algorithm used in the Internet congestion control to EV charging using only local node measurements. We analyze the relationship between distance and grid voltage, and show how to extract this information from local measurements. Then, we present a detailed analysis to understand the relationship between distance and charging power in a distribution network to better address the fairness in the proposed AIMD EV charging algorithm. Results show that localized information at charging node voltages include important signature information on grid congestion and can be used to implement AIMD control for EV charging.

Network utility maximization revisited: Three issues and their resolution

Distributed and iterative network utility maximization algorithms, such as the primal–dual algorithms or the network-user decomposition algorithms, often involve trajectories where the iterates may be infeasible, convergence to the optimal points of relaxed problems different from the original, or convergence to local maxima. In this paper, we highlight the three issues with iterative algorithms. We then propose a distributed and iterative algorithm that does not suffer from the three issues. In particular, we assert the feasibility of the algorithm’s iterates at all times, convergence to global maximum of the given problem (rather than to global maximum of a relaxed problem), and avoidance of any associated spurious rest points of the dynamics. A benchmark algorithm due to Kelly et al. (1998) involves fast user updates coupled with slow network updates in the form of additive-increase multiplicative-decrease of suggested user flows. The proposed algorithm may be viewed as one with fast user updates and fast network updates that keeps the iterates feasible at all times. Simulations suggest that the convergence rate of the ordinary differential equation (ODE) tracked by our proposed algorithm’s iterates is comparable to that of the ODE for the aforementioned benchmark algorithm.

Comparative analysis of cumulative distribution function for TCP congestion window size and triple‐duplicate period

SummaryVarious version of Transmission Control Protocol (TCP) congestion window model are available in the literature. TCP congestion control is handled in two phases: (a) slow start and (b) congestion avoidance. This paper deals with congestion avoidance phase that is based on additive increase multiplicative decrease (AIMD) mechanism in which window size either increase by one or cuts to half of the previous window size. For distribution of window sizes, a model developed by Yan et al has been presented for two loss indications: (a) triple duplicate (TD) and (b) time‐outs (TOs) only while it does not deal with the window size limitation. It does not allow to cap over the TCP congestion window size if the buffer size is restricted to a fixed value though window size could not move beyond that. The unconstrained window size moves on in the presence of low loss probability. In this paper, we propose a model for the case where the window size is bounded by a maximum value that makes applicable to window size limitation as well. Further, the work has been extended to develop a new model that obtains a cumulative distribution function for TD periods (TDPs). The proposed model is validated on ns‐2, and we conclude that observed results for distribution function are very closed to our proposed model.