Search

In this paper, we introduce Deep Reinforcement Learning (DRL) for congestion control in the Transmission Control Protocol/Internet Protocol (TCP/IP) networks. We propose a weighted ensemble DRL model that combines four DRL models Deep Q-Learning (DQN), Proximal Policy Optimisation (PPO), Deep Deterministic Policy Gradient (DDPG), and Twin Delay DDPG (rlTD3). These four models are designed with varying action spaces for efficient congestion control in an Active Queue Management (AQM) system. The proposed model outperformed single DRL models and established congestion control algorithms like Random Early Detection (RED) in normal and stress testing. The proposed model improved throughput by 4% and delays by 10.52% compared to DQN and 2.92% compared to DDPG. The proposed model has shown promising results in managing congestion in dynamic network environments and handling high-traffic loads. [ABSTRACT FROM AUTHOR]

Abstract Wireless sensor networks (WSN) may be composed by tens or hundreds of nodes. Nodes measure variables and send their values to a base station through a shared low data rate wireless communication channel. Communication protocols for WSNs have been developed in order to support a large number of nodes. However, if a great deal of data is transmitted through a low speed communication channel, data packet congestion may happen. Congestion may lead to high latencies or data loss. Data congestion could take place in WSNs with a large number of nodes sending data periodically, or in WSNs where the data rate can increase when a specific event is detected. In this paper we analyze the conditions that lead to congestion of the communication channel considering the following nodes configuration parameters: (1) transmission time intervals, (2) data packet generation rate, and (3) transmitter output power level. [ABSTRACT FROM AUTHOR]

Congestion in wireless networks is one of the major causes of system inefficiency, and with router load being the main contributor to overall network traffic flow, congestion is very dependent on the level of router load and how it can be effectively managed. This paper presents a novel low-complexity successive load shifting (SLS) technique for intelligently shifting router load between network routers by predicting the probability of congestion occurrences in the network and exploiting the topology to reassign load to minimise congestion. Crucially, SLS does not compromise the data rate in avoiding congestion and is able to be seamlessly embedded into existing protocols with only a small increase incurred in system overheads. The performance of SLS has been extensively tested and critically evaluated using the widely adopted TCP and UDP protocols, with results confirming both significant throughput gains and superior packet loss performance. [ABSTRACT FROM AUTHOR]

In this paper, we propose a client-based solution to detect “evil twin” attacks in wireless local area networks (WLANs). An evil twin is a kind of rogue Wi-Fi access point (AP) which has the same SSID name as a legitimate one and is set up by an attacker. After a victim associates his device with an evil twin, an attacker can eavesdrop sensitive data forwarded through the evil twin. Most existing detection solutions are administrator-based, which are used by wireless network administrators to verify whether a given AP is in an authorized list or not. Such administrator-based solutions are limited, hardly maintained, and difficult to protect users 24–7. Hence, we propose a client-based detection mechanism, called evil twin detector, to detect this type of attacks. An evil twin detector changes its wireless network interface card (WNIC) to monitor mode to capture wireless TCP/IP packets. Through analyzing captured packets, our detector allows client users to easily and precisely detect an evil twin, thus avoids threats created by evil twins. Our method does not need to know any authorized AP list, and does not rely on data training or machine learning technique. Finally, we implement a detecting system on Windows 7. [ABSTRACT FROM AUTHOR]

SYN flooding attack is a threat that has been designed based on vulnerabilities of the connection establishment phase of the TCP protocol. In this attack some sources send a large number of TCP SYN segments, without completing the third handshake step to quickly exhaust connection resources of the victim server. Hence, a main part of the server's buffer space is allocated to the attack half open connections and incoming new connection requests will be blocked. This paper proposes a novel framework, in which, the defense issue is formulated as an optimization problem. Then it employs the particle swarm optimization (PSO) algorithm to solve this optimization problem. Our theoretical analysis and packet-level simulations in ns-2 environment show that the proposed defense strategy called PSO_SYN decreases the number of blocked TCP connection requests and cuts down share of attack connections from the buffer space. [ABSTRACT FROM AUTHOR]

IPv6 protocol, which should replace the actual IPv4 protocol, brings many new possibilities and improvements considering simplicity, routing speed, quality of service and security. In comparison to IPv4, IPv6 improves mechanisms for assuring a secure and confidential transfer of information. Despite these improvements, network security remains a very important issue since there are some security threats and attack types that can affect IPv6 network. This paper deals with security issues in IPv6 networks. Security improvements and extensions in the IPv6 protocol are described and explained. Also, security comparison to IPv4 is made. A description of the experimental IPv6 network and a description of tools used for security testing are presented in the paper. Security threats similar in IPv4 and IPv6 networks are described, and some security issues specific for IPv6 networks are also analysed. Different types of attacks in IPv6 networks are analysed and some suggestions for their avoidance are given. Considering security, especially problematic is the transition period of coexistence of both protocols. Because of that, security issues due to different transition mechanisms are analysed. Further, the paper studies firewalls in IPv6 networks. Implementation of firewalls in IPv6 networks and IPv6 specific firewall configurations are analysed. Different tests of firewalls are performed, and their results are analysed. Also, comparison with IPv4 firewalls is made. Some suggestions referring to proper deployment of firewalls are given. This paper also deals with detection of unauthorised intrusion. Different approaches to intrusion detection are explained and different types of intrusion detection systems are described. Suggestions for proper positioning of intrusion detection systems in the local area network are given. In absence of non-commercial intrusion detection systems with IPv6 support, some alternative possibilities of intrusion detection are explained. The paper analyses methods of intrusion detection by using tools for network traffic capturing and analysis (with IPv6 support). Different types of attacks are performed and their effects are presented and explained. Instructions for recognition and detection of different attacks are given. Some recommendations for avoiding certain attack types or reducing their effect are given. Practical advices and guidelines in implementation of security mechanisms for packet filtering and detection of unauthorized intrusion are emphasized. Finally, some recommendations for improving security mechanisms and guidelines for further development of intrusion detection systems with IPv6 support are given. Keywords: IPv6; Network security; Firewall; Intrusion detection

Abstract: Active queue management has been proved to be an effective mechanism at the intermediate nodes to alleviate network congestion. A novel robust active queue management scheme based on H-infinity feedback control theory is proposed in this paper. A TCP model with link capacity disturbances, modeling uncertainties and time-varying delay is investigated, and the objective is to design a feedback controller which guarantee the queue length stability and robustness against the external disturbances and model perturbations. Meanwhile a simple on-line estimation of TCP window size is adopted instead of state observer which reduces computation burden significantly. The controller parameters are obtained through the standard linear matrix inequality techniques. Finally NS-2 simulation results demonstrate effectiveness of the proposed method. [Copyright &y& Elsevier]

Abstract: In this paper, we present a resource-aware and quality-fair video content sharing system. When a video sharing server has insufficient uplink bandwidth and needs to serve multiple video content sharing services via streaming or downloading to other client peers using TCP transport, each service shares the limited uplink bandwidth equitably, due to the fair sharing characteristics inherent in TCP. However this bandwidth fair sharing cannot always guarantee quality fairness among the services, due to the specific requirements for video-streaming services, such as the playout rate and the size of the playout buffer. In our system, the server uses multiple TCP connections adaptively, depending on the anticipated status of each client playout buffer, to guarantee the bandwidth of each video-streaming session. By guaranteeing the quality of each video-streaming session, without the quality loss of other service sessions, the proposed system can successfully achieve service quality fairness. Simulation results show that our proposed algorithm can dramatically enhance the quality of each streaming session and thus provide service quality fairness among simultaneous multiple heterogeneous video-streaming services and content download services. [Copyright &y& Elsevier]

Abstract: This paper addresses the problem of parallel dynamic security assessment applications from static homogeneous cluster environment to dynamic heterogeneous grid environment. Functional parallelism and data parallelism are supported by each of the message passing interface model and TCP/IP model. To consider the differences in heterogeneous computing resources and complexity of large-scale power system communities, a kernel-based multilevel algorithm is proposed for network partitioning. Since the bottleneck in distributed computation is low speed network communication, a bi-level latency exploitation technique is introduced for numerically solving system differential equations. The proposed grid-based implementation includes the core simulation engine, grid computing middleware, a Python interface and Python front-end utilities. Tests for a 39-bus network, a 4000-bus network and a 10,000-bus network are reported, and the results of these experiments demonstrate that the proposed scheme is able to execute the distributed simulations on computational grid infrastructure and provide efficient parallelism. [Copyright &y& Elsevier]

Abstract: There has been a surge of interest for multimedia applications over wireless networks in recent years. A colossal number of ways have been proposed to decrease delay, delay jitter and loss in wireless networks and good user-perceived quality in video over internet. This paper studies the multimedia over heterogeneous wireless networks, requirements and problems, and proposes a new scheme to overcome the obstacles. The proposed scheme, takes into account the effects of Application-Level Wireless Multilevel ECN marking (AWMECN), thus helps us overcome the congestion/loss mistake problems. For handoff, handover and lossy link problems, it is considered that a freezing mechanism is in use in application layer and assumed that the upper layers can be aware of disconnection periods to make the rate adaptation decisions. Also a new scheme has been added to receiver to gracefully degrade the quality when no other action is available to combat the long delays without data which is caused by handoffs and wireless temporary disconnections. The transport layer mechanism is chosen to be UDP for avoiding TCP regarding problems. We believe that obtaining a good quality of video depends on good performance of all layers and tried to use the best mechanism in each layer. [Copyright &y& Elsevier]

Abstract: Active Queue Management (AQM) is a proven strategy to efficiently maintain queues and ensure high utilization of Transmission Control Protocol (TCP) network resources. The fundamental mechanism is to manage incoming packet rates at a router to prevent incipient network congestion. In this paper, we present an efficient neural network AQM system as a queue controller. The recurrent neural network has a Multi-layer Perceptron-Infinite Impulse Response (MLP-IIR) structure. Three distinct neural AQMs are trained under different network scenarios involving traffic levels. Selecting one of three neural AQMs is based on posterior probability history of traffic level. In addition, we investigate stochastic modeling of the network dynamics by a Dynamic Bayesian Network (DBN). This model allows implementation of a predictive AQM system in which queue dynamics are predicted and used for error prediction via online DBN estimation. Our AQM method is evaluated through simulation experiments both using an Ordinary Differential Equation (ODE) network model and using OPNET©. The simulation results demonstrate that our adaptive neural AQM outperforms Random Early Detection (RED) and Proportional-Integral-Derivative (PID) based AQM. [Copyright &y& Elsevier]

To link to full-text access for this article, visit this link: http://dx.doi.org/10.1016/j.compeleceng.2008.05.006 Byline: Wei Zhang (a), Liansheng Tan (a)(b), Gang Peng (a) Keywords: Random early detection (RED); Active queue management (AQM); Stability; Hermite matrix; Congestion control Abstract: One main TCP congestion control objective is, by dynamically adjusting the source window size according to the router queue level, to stabilize the buffer queue length at a given target, thereby achieving predictable queueing delay, reducing packet loss and maximizing link utilization. One difficulty therein is the TCP acknowledging actions will experience a time delay from the router to the source in a TCP system. In this paper, a time-delay control theory is applied to analyze the mechanism of packet-dropping at router and the window-updating in TCP source in TCP congestion control for a TCP/RED dynamic model. We then derive explicit conditions under which the TCP/RED system is asymptotically stable in terms of the instantaneous queue. We discuss the convergence of the buffer queue lengths in the routers. Our results suggest that, if the network parameters satisfy certain conditions, the TCP/RED system is stable and its queue length can converge to any target. We illustrate the theoretical results using ns2 simulations and demonstrate that the network can achieve good performance and converge to the arbitrary target queues. Author Affiliation: (a) Department of Computer Science, Central China Normal University, Wuhan 430079, PR China (b) Research School of Information Sciences and Engineering, The Australian National University, Canberra ACT 0200, Australia Article History: Received 12 June 2007; Revised 27 April 2008; Accepted 8 May 2008