Measurement and performance of a cognitive packet network

doi:10.1016/S1389-1286(01)00253-5

Computer Networks

Volume 37, Issue 6, December 2001, Pages 691-701

https://doi.org/10.1016/S1389-1286(01)00253-5 Get rights and content

Abstract

As the size of the Internet grows by orders of magnitude both in terms of users, number of IP addresses, and number of routers, and as the links we use (be they wired, optical or wireless) continuously evolve and provide varying reliability and quality of service, the IP based network architecture that we know so well will have to evolve and change. Both scalability and QoS have become key issues. We are currently conducting a research project that revisits the IP routing architecture issues and proposes new designs for routers. As part of this effort, this paper discusses a packet network architecture called a cognitive packet network (CPN), in which intelligent capabilities for routing and flow control are moved towards the packets, rather than being concentrated in the nodes. In this paper we outline the design of the CPN architecture, and discuss the quality-of-service based routing algorithm that we have designed and implemented. We then present our test-bed and report on extensive measurement experiments that we have conducted.

Introduction

In several recent papers [1], [2], [3] we have proposed a new network architecture called “cognitive packet networks” (CPNs). These are store-and-forward packet networks in which intelligence is built into the packets, rather than at the routers or in the high-level protocols. CPNs carry three major types of packets: smart packets (SPs), dumb packets (DPs) and acknowledgments (ACK). Smart or cognitive packets (CPs) route themselves, they learn to avoid congestion and to avoid being lost or destroyed. They learn from their own observations about the network and from the experience of other packets. They rely minimally on routers. When a SP arrives at a destination, an acknowledgment packet (AP) is generated by the destination and the ACK heads back to the source of the SP along the reverse route. As it traverses successive routers, it is used to update mailboxes (MBs) in the CPN routers, and when it reaches the source node it provides source routing information for dumb packets. Dumb CPN packets of a specific QoS class use successful routes which have been selected in this manner by the SPs of that class.

This paper reviews the basic concepts of CPNs. Then, we present analytical results for best and worst case performance. We then describe a test-bed network we have designed and implemented in order to demonstrate these ideas. We provide measurement data on the test-bed to illustrate the capacity of the network to adapt to changes in traffic load and to failures of links. Finally we report measurements which evaluate the impact of the ratio of smart to dumb packets on the end-to-end delay experienced by all of the packets.

Section snippets

Cognitive packets and cognitive packet networks

Learning algorithms and adaptation have been suggested for telecommunication systems in the past [4], [6]. However these concepts have not been fully exploited in networks because of the lack of an adequate framework allowing decentralized control of communications. In CPN, smart packets serve as “explorers” for different source–destination (S–D) pairs; they rely minimally on routers, so that network nodes only serve as buffers, MBs and processors. We use the term cognitive packet (CP) and

Routing algorithm using reinforcement learning

We use random neural networks (RNN) with reinforcement learning (RNNRL) in order to implement the SP routing algorithm. A recurrent RNN is used both for storing the CM and making decisions. The weights of the RNN are updated so that decisions are reinforced or weakened depending on how they have been observed to contribute to the success of the QoS goal. Our earlier experience with simulations of CPN [3], [7] as well as our current test-bed implementation have shown the practical validity of

An analytical model of smart packets in cognitive packet network

We have examined a simple mathematical model to predict the end-to-end delay experienced by SPs. The worst case performance is obtained by considering SPs which try to find the route to their destination by moving at random, with two constraints related to the topology of the network which will be discussed below. The network topology that we use to evaluate the worst case is identical to the one which we have used in our simulations, except that it has a cylindrical topology with “R” circles

An experimental cognitive packet network test-bed: protocol design and software implementation

In this section we describe aspects of the CPN protocol design and details of a test-bed implementation. The software has been integrated into the LINUX kernel 2.2.x. with minimal changes in the existing networking code and is independent of the physical transport technology. The network interface is compatible with the popular BSD4.3 socket layer in LINUX, and provides a single application program interface (API) for the programmer to access the CPN protocol. The LINUX kernel support for low

Network measurements

The purpose of the measurements we describe on the test-bed is to evaluate the CPN architecture with respect to its dynamic behavior and ability to adapt to changing network conditions. These conditions include both changes in traffic patterns and possible failures in links. With respect to the network topology described in Fig. 2, we have conducted measurements for a main flow of traffic from Node 10 (left) to Node 5 (right). For the first set of experiments, the input rate was fixed to five

Conclusions

CPNs are a new packet network paradigm which address some of the needs of global networking. CPN simplifies router architecture by transfering the control of QoS based best-effort routing to the packets, away from the routers. Routing tables are replaced by reinforcement-algorithm based routing functions. A CPN carries three distinct types of packets: Smart or cognitive packets which search for routes based on a QoS driven reinforcement learning algorithm, ACKPs which bring back route

References (7)

K.S. Narendra et al.
The use of learning algorithms in telephone traffic routing––a methodology
Automatica
(1983)
E. Gelenbe, E. Seref, Z. Xu, Towards networks with intelligent packets, Proceedings of the IEEE-ICTAI Conference on...
E. Gelenbe, R. Lent, Z. Xu, Towards networks with cognitive packets, Opening Key-Note Paper, Proceedings of the IEEE...

There are more references available in the full text version of this article.

Cited by (101)

A survey on cognitive packet networks: Taxonomy, state-of-the-art, recurrent neural networks, and QoS metrics
2022, Journal of King Saud University - Computer and Information Sciences
Resilience and flexibility are the means by which a network stands for an efficient service provider platform. Cognitive packet network (CPN) is such a paradigm that solely depends on the behavioral and intelligence capacity of the packets that carry valuable information within the network. Unlike conventional networks, CPN relies on the decision-making ability of the packets. In CPN, different types of packets take part in finding the optimal route from the source to destination. Further, packets dynamically adapt environmental changes by using recurrent neural network-based learning algorithms. Mapping cognitive features along with enrichment of quality of service is the main task of CPN. Due to continuous updates in the packet information vulnerable elements are found to be less prone to attack the CPN, thus resulting in a robust networking solution. In this paper we firstly present a novel taxonomic structure of the CPN. Then, we discuss the stringent routing techniques normally used in CPN. We also investigate various adaptive aware features in CPN. Next, we present intuitive emergency service and sensor network design prospects of the CPN. The article also discusses various means of cognitive aware mitigation aligned with the packets in CPN. Finally, we deliver how CPN is secure and available security aspects.
Deep Reinforcement Learning with the Random Neural Network
2022, Engineering Applications of Artificial Intelligence
This paper proposes a Deep Reinforcement Learning (DRL) algorithm that expands the Random Neural Network (RNN) Reinforcement Learning (RL) method to include the previous learnings entirely from previous rewards, rather than only the actual one. The Random Neural Network weighs are updated with the current reward and the previous values, including time and memory. This addition makes DRL slower to make decisions, although it also increases its performance in some experiments. Several configurations to introduce DRL, such as sampling rate and memory duration, are also proposed and analysed in this article. The proposed DRL algorithm is included in a decision process that predicts trends: upward, downward and equal market directions in addition to values. Experimental results based on market prices demonstrate that the addition of Deep Learning to the Reinforcement Learning algorithm increases its performance slightly in some experiments; however, it also increases its computational cost. In random environments such as the stock market, it is preferable to make decisions based on the previous values (short memory) rather than historical records (long memory).
Deep Learning Clusters in the Cognitive Packet Network
2020, Neurocomputing
Citation Excerpt :
Data networks collect information from users and transmit it to different locations; to perform this activity, they are required to make routing decisions based on different Quality of Service metrics while storing routing tables in memory under the threat of Cyber attacks. This paper proposes the Cognitive Packet Network (CPN) [11–15] with an additional Deep Learning (DL) cluster [31,32] structure that emulates how the brain operates. The proposed model adds a layer of specialised Deep Learning management clusters that take the final routing decision; DL clusters behave as a long term memory to remember network identity: QoS metrics and Cyber keys.
The Cognitive Packet Network (CPN) bases its routing decisions and flow control on the Random Neural Network (RNN) Reinforcement Learning algorithm; this paper proposes the addition of a Deep Learning (DL) Cluster management structure to the CPN for Quality of Service metrics (Delay Loss and Bandwidth), Cyber Security keys (User, Packet and Node) and Management decisions (QoS, Cyber and CEO). The RNN already models how neurons transmit information using positive and negative impulsive signals whereas the proposed additional Deep Learning structure emulates the way the brain learns and takes decisions; this paper presents a brain model as the combination of both learning algorithms, RNN and DL. The proposed model has been simulated under different network sizes and scenarios and it has been validated against the CPN itself without DL clusters. The simulation results are promising; the presented CPN with DL clusters as a mechanism to transmit, learn and make packet routing decisions is a step closer to emulate the way the brain transmits information, learns the environment and takes decisions.
EROL GELENBE: A CAREER in MULTI-DISCIPLINARY PROBABILITY MODELS
2016, Probability in the Engineering and Informational Sciences
Review of Some Recent European Cybersecurity Research and Innovation Projects
2022, Infocommunications Journal
AI and Quality of Service Driven Attack Detection, Mitigation and Energy Optimization: A Review of Some EU Project Results
2022, Communications in Computer and Information Science

View all citing articles on Scopus

Erol Gelenbe is the author of several books in the area of queuing networks and computer system performance which have appeared in English, French, Japanese and Korean. He has graduated some 55 Ph.D. students and published over 100 articles in leading journals of Computer Science, Electrical Engineering and applied Probability. He currently works on novel network architectures and protocols, as well as on multimedia applications. His recent papers have appeared in the Proceedings of the IEEE, the IEEE Journal on Selected Areas in Communications, the IEEE Transactions on Neural Networks and the journal Performance Evaluation. He holds the University Chair of Electrical Engineering and Computer Science (EECS) at the University of Central Florida, where he is Director of the School of EECS and Associate Dean of Engineering. He holds a Ph.D. from Polytechnic University (Brooklyn, New York), and the Doctor of Science degree from the University Pierre et Marie Curie (Paris VI) in France. His honors include the “Grand Prix France Telecom” of the French Academy, the IFIP Silver Core, the Chevalier du Mérite of France, an Honorary Doctorate from the University of Rome II, and Fellow of the IEEE. His research is funded by NSF, the US Army and industry.

Ricardo Lent was born in Chiclayo, Peru. He received the B.S. degree and the “Ingeniero Electronico” title from Universidad Ricardo Palma, Lima, Peru, in 1992, and the M.S. degree in Electrical Engineering from the Universidad Nacional de Ingenieria, Lima, Peru in 1997. From 1992 to 1999 he made major contributions toward the establishment of the first Internet network in Peru. He has also been a Lecturer at the Universidad Nacional de Ingenieria and at the Universidad Nacional Pedro Ruiz Gallo in Peru. He has co-authored several papers in IEEE and other international conferences and is also co-author of a paper in the journal Performance Evaluation. He is currently pursuing a doctoral degree in Computer Science at the University of Central Florida, Orlando, FL.

Zhiguang Xu received the undergraduate degree in Computer Engineering from Beijing University of Posts and Telecommunications, China and the M.S. degree in Computer Science from University of Central Florida. Since January 1999, he has been a Graduate Student working towards the Ph.D. degree in Computer Science at the University of Central Florida. He was with Nortel Networks for four years. His research covers neural-network applications in modelling and simulation and in the design of novel packet switching architectures. He recently co-authored several conference papers, which have appeared in the journals Proceedings of the IEEE and Performance Evaluation. The IEEE Conference on Tools for Artificial Intelligence in 1999, the International Symposium on Computer and Information Sciences in 1999, and the IEEE MASCOTS Workshop 2000 in San Francisco.

^☆: The research reported in this paper was supported by US Army Simulation and Training Command and by Giganet Technologies, Inc.

View full text

Measurement and performance of a cognitive packet network☆