Generalizable and Interpretable Deep Learning for Network Congestion Prediction

2021 ◽  
Author(s):  
Konstantinos Poularakis ◽  
Qiaofeng Qin ◽  
Franck Le ◽  
Sastry Kompella ◽  
Leandros Tassiulas
Keyword(s):  
Deep Learning ◽  
Network Congestion ◽  
Congestion Prediction

Deep Learning Based Congestion Prediction Using PROBE Trajectory Data

CICTP 2019 ◽  
2019 ◽  
Author(s):  
Jingqiu Guo ◽  
Yangzexi Liu ◽  
Yibing Wang ◽  
Ken Yang
Keyword(s):  
Deep Learning ◽  
Trajectory Data ◽  
Congestion Prediction

Region-wide congestion prediction and control using deep learning

2020 ◽  
Vol 116 ◽  
pp. 102624
Author(s):  
Sudatta Mohanty ◽  
Alexey Pozdnukhov ◽  
Michael Cassidy
Keyword(s):  
Deep Learning ◽  
Congestion Prediction ◽  
Prediction And Control ◽  
And Control

scholarly journals Proactive Congestion Avoidance for Distributed Deep Learning

Sensors ◽  
2020 ◽  
Vol 21 (1) ◽  
pp. 174
Author(s):  
Minkoo Kang ◽  
Gyeongsik Yang ◽  
Yeonho Yoo ◽  
Chuck Yoo
Keyword(s):  
Neural Network ◽  
Deep Learning ◽  
Queue Length ◽  
Deep Neural Network ◽  
Congestion Avoidance ◽  
Model Parameters ◽  
Neural Network Training ◽  
Network Congestion ◽  
Network Training ◽  
Deep Learning Model

This paper presents “Proactive Congestion Notification” (PCN), a congestion-avoidance technique for distributed deep learning (DDL). DDL is widely used to scale out and accelerate deep neural network training. In DDL, each worker trains a copy of the deep learning model with different training inputs and synchronizes the model gradients at the end of each iteration. However, it is well known that the network communication for synchronizing model parameters is the main bottleneck in DDL. Our key observation is that the DDL architecture makes each worker generate burst traffic every iteration, which causes network congestion and in turn degrades the throughput of DDL traffic. Based on this observation, the key idea behind PCN is to prevent potential congestion by proactively regulating the switch queue length before DDL burst traffic arrives at the switch, which prepares the switches for handling incoming DDL bursts. In our evaluation, PCN improves the throughput of DDL traffic by 72% on average.

scholarly journals Deep Learning Entrusted to Fog Nodes (DLEFN) Based Smart Agriculture

2020 ◽  
Vol 10 (4) ◽  
pp. 1544 ◽  
Author(s):  
Kyuchang Lee ◽  
Bhagya Nathali Silva ◽  
Kijun Han
Keyword(s):  
Cloud Computing ◽  
Deep Learning ◽  
Fog Computing ◽  
Multimedia Data ◽  
Network Congestion ◽  
Service Delay ◽  
Comparison Results ◽  
Smart Agriculture ◽  
Iot Devices ◽  
Computing Capacity

Colossal amounts of unstructured multimedia data are generated in the modern Internet of Things (IoT) environment. Nowadays, deep learning (DL) techniques are utilized to extract useful information from the data that are generated constantly. Nevertheless, integrating DL methods with IoT devices is a challenging issue due to their restricted computational capacity. Although cloud computing solves this issue, it has some problems such as service delay and network congestion. Hence, fog computing has emerged as a breakthrough way to solve the problems of using cloud computing. In this article, we propose a strategy that assigns a portion of the DL layers to fog nodes in a fog-computing-based smart agriculture environment. The proposed deep learning entrusted to fog nodes (DLEFN) algorithm decides the optimal layers of DL model to execute on each fog node, considering their available computing capacity and bandwidth. The DLEFN individually calculates the optimal layers for each fog node with dissimilar computational capacities and bandwidth. In a similar experimental environment, comparison results clearly showed that proposed method accommodated more DL application than other existing assignment methods and utilized resources efficiently while reducing network congestion and processing burden on the cloud.

scholarly journals Large-Scale Transportation Network Congestion Evolution Prediction Using Deep Learning Theory

PLoS ONE ◽  
2015 ◽  
Vol 10 (3) ◽  
pp. e0119044 ◽  
Author(s):  
Xiaolei Ma ◽  
Haiyang Yu ◽  
Yunpeng Wang ◽  
Yinhai Wang
Keyword(s):  
Deep Learning ◽  
Learning Theory ◽  
Large Scale ◽  
Transportation Network ◽  
Network Congestion

scholarly journals NETWORK CONGESTION PREDICTION USING CLASSIFICATION

2021 ◽  
pp. 65-67
Author(s):  
A Prasanth Rao ◽  
Akula Venkata Sai Akhil ◽  
Akku Vivek ◽  
Preetham Reddy Yaramada
Keyword(s):  
Packet Loss ◽  
System Performance ◽  
Network Performance ◽  
Threshold Value ◽  
Transport Layer ◽  
Classification Algorithms ◽  
Network Congestion ◽  
Data Loss ◽  
Loss And Damage ◽  
Congestion Prediction

Network congestion is a situation that happens when a network is un- able to handle data is more than its threshold value when load becomes high. This situation is known to be network congestion which causes packets to be dropped on the network due to overflow of buffer and therefore leads to data loss and unreliable connection. Therefore, effective congestion control is an important is- sue that needs to be addressed in the transport layer. There are various factors such as hardware, software and miscellaneous factors can lead to network con- gestion. Hardware factors include noncompatible hardware, outdated routers and too many devices connecting to a single router. Software factors include some of the devices following their own protocol, ineffective communication protocols, improper fire walls and proper limitations not defined on the file sizes to be transferred. Poor network design, network hacking and over subscription leads to the miscellaneous factors. As a consequence of such factors, network performance will degrade dramatically, and system performance will be affected. This is an undesirable condition that needs to be corrected. Thus, our model uses classification algorithms which help in predicting network congestion before- hand, thuspreventing the packet loss and damage.

Sign in / Sign up

Export Citation Format

Share Document