Adaptive AIMD Congestion Control

Kesselman, Alex; Mansour, Yishay

doi:10.1007/s00453-005-1160-3

Cited by 9 publications

(3 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Classical approaches that have been considered are MIMD (Multiplicative Increase, Multiplicative Decrease [Kel03]) and AIMD (Additive Increase, Multiplicative Decrease [CJ89]). Many other researchers studied congestion control for the AIMD model, which resulted in various extensions of the original work, see for example [CS12], [KM05], [LT05]. Although these protocols work for arbitrary initial probabilities, their auxiliary variables are always assumed to be well-initialized.…”

Section: Related Workmentioning

confidence: 99%

A Loosely Self-stabilizing Protocol for Randomized Congestion Control with Logarithmic Memory

Feldmann

Götte

Scheideler

2019

Lecture Notes in Computer Science

View full text Add to dashboard Cite

We consider congestion control in peer-to-peer distributed systems. The problem can be reduced to the following scenario: Consider a set V of n peers (called clients in this paper) that want to send messages to a fixed common peer (called server in this paper). We assume that each client v ∈ V sends a message with probability p(v) ∈ [0, 1) and the server has a capacity of σ ∈ N, i.e., it can recieve at most σ messages per round and excess messages are dropped. The server can modify these probabilities when clients send messages. Ideally, we wish to converge to a state with p(v) = σ and p(v) = p(w) for all v, w ∈ V . We propose a loosely self-stabilizing protocol with a slightly relaxed legitimate state. Our protocol lets the system converge from any initial state to a state where. This property is then maintained for Ω(n c ) rounds in expectation. In particular, the initial client probabilities and server variables are not necessarily well-defined, i.e., they may have arbitrary values.Our protocol uses only O(W + log n) bits of memory where W is length of node identifiers, making it very lightweight. Finally we state a lower bound on the convergence time an see that our protocol performs asymptotically optimal (up to some polylogarithmic factor). * This is an extended version of a paper which will appear in SSS 2019. This work was partially supported by the German Research Foundation (DFG) within the Collaborative Research Center On-The-Fly Computing (GZ: SFB 901/3) under the project number 160364472.Consider a set of n nodes (called clients in this paper) that want to continuously send messages to a fixed node (called server) with a certain probability in each round. The server is not aware of its connections and has limited capabilities with regard to the number of messages it is able to receive in each round and its internal memory. The task for the server is to use a congestion control protocol to modify the client probabilities such that the server receives only a constant amount of messages in each round (on expectation). As client probabilities may be arbitrary at the beginning, we further require the protocol to be self-stabilizing, i.e., it should be able to reach its goal starting from any arbitrary initial state. Self-stabilization comes with the advantage that the protocol is able to recover from transient faults like message loss or blackout of processes automatically. As the system grows larger, these kinds of faults occur more often, which makes self-stabilization as a concept very desirable.At first glance, one may think that this setting only applies to client/server-architectures. However, we believe that solving this problem is quite important for distributed systems where nodes constantly have to communicate with their neighbors. Also there are distributed systems where nodes are not aware of their incoming connections, e.g. in rooted trees, random graphs [MS06] or linearized de Bruijn networks [RSS11]. On these networks one is able to effectively perform many important techniques relevant to ...

show abstract

Section: Related Workmentioning

confidence: 99%

A Loosely Self-stabilizing Protocol for Randomized Congestion Control with Logarithmic Memory

Feldmann

Götte

Scheideler

2019

Lecture Notes in Computer Science

View full text Add to dashboard Cite

show abstract

“…TCP congestion control generally contains four parts which are slow-start, congestion avoidance, fast retransmission and fast recovery [6]. After TCP connection is established, slow-start algorithm begins to execute firstly.…”

Section: Tcp Congestion Controlmentioning

confidence: 99%

Applying the Linear Neural Network to TCP Congestion Control

Niu¹

2015

Proceedings of the 5th International Conference on Advanced Design and Manufacturing Engineering

View full text Add to dashboard Cite

Abstract. The previous TCP protocol cannot predict congestion. Only when the sender receives more than three acknowledgements or the retransmission timer is out can it realize that congestion has occurred. We train the linear neural network by using round-trip time and current TCP throughput as its inputs. As a result, we get the decision boundary, which could predict whether the current network is in congestion or not. Simulation results show that, when applied to TCP congestion control, it can effectively predict the occurrence of congestion, so congestion window could make adjustments as soon as possible to reduce the probability of congestion collapse.

show abstract

“…Exceptions are the studies of discrete-time dynamics in [8][9][10][11][12] and simulation studies [13]. Proposals of discrete-time adaptive AIMD congestion control have also been made [10,14,15], adopting approaches based on increasing fairness. More recent work in the area of decentralized resource allocation has looked at modified resource distribution fairness measures [16] and game-theoretic-based/optimization-based approaches, using utility functions for each user (see [17,18] and the references therein).…”

mentioning

confidence: 99%

A decentralized adaptive algorithm for fair allocation of a shared resource

Bhaya

Kaszkurewicz

2015

Adaptive Control & Signal

View full text Add to dashboard Cite

This paper views the classical Chiu-Jain algorithm, originally proposed for congestion control of network links, as a decentralized algorithm for the fair allocation of a total of c units of a shared resource among n users. A new analysis is given of the general case of additive increase and multiplicative decrease (AIMD) dynamics, from the perspective of virtual equilibria and variable structure systems, leading to a better understanding of the Chiu-Jain algorithm, which is one example of AIMD dynamics. It is shown that the variable structure discrete dynamical system that describes the evolution of the share of each individual, starting from an arbitrary initial allocation, always attains a neighbourhood of the fair share (c=n) for each user, under the assumption that the latter is known. Subsequently, a new adaptive version of the algorithm, called adaptive AIMD, is described, with the same property of converging to the fair share, without assuming that it is known. Simulations that show the behaviour and advantages of the proposed adaptive AIMD adaptive algorithm are given.

show abstract

Adaptive AIMD Congestion Control

Cited by 9 publications

References 27 publications

A Loosely Self-stabilizing Protocol for Randomized Congestion Control with Logarithmic Memory

A Loosely Self-stabilizing Protocol for Randomized Congestion Control with Logarithmic Memory

Applying the Linear Neural Network to TCP Congestion Control

A decentralized adaptive algorithm for fair allocation of a shared resource

Contact Info

Product

Resources

About