Networking Projects at UNCG

Details:

Increasing packet loss rates and low throughput caused by an exponential increase in computer network traffic are the two major concerns in the highly congested Internet today. The Random Early Detection (RED), an active queue management technique, has been suggested as a solution to these problems [1]. However, it does not provide an acceptable solution because of its dependency on a number of control parameters such as the weight used in exponential weighted averaging (wq), the minimum and the maximum thresholds used for queue management (th_min, th_max) and the maximum packet-dropping probability (max_p), and unpredictable system parameters such as the round-trip time of network connections and the load of the network (flows or connections). We will develop a technique using Markov model transition probabilities of the queue so that it can adaptively change the control parameters based on the queue dynamics. This technique will detect the congestion very early and adjust the packet-dropping probability so that RED can make wise packet-dropping decisions and keep the average queue size closer to the target queue size. The RED algorithm is recommended by the Internet Engineering Task Force (IETF) for use in the routers of next generation [2], therefore, our goal in this project is to suggest a solution with minimal changes to the overall RED algorithm. The proposed scheme will be designed to perform well in the real network conditions and it will be illustrated using simulation with Network Simulator (NS-2) software.

Transmission Control Protocol (TCP) plays a major role in reliable transmission of data over wired and wireless networks. The nature of a shared wireless medium and dynamic location dependency causes TCP unfairness in ad hoc wireless networks. If we view a wireless node and its interfering nodes to form a "neighborhood", the aggregate of local queues at these nodes represents the distributed queue for this neighborhood. However, this queue is not a First-In-First-Out (FIFO) queue because the traffic flows sharing the queue have different, dynamically changing priorities determined by the topology and traffic patterns. Thus, they get different feedback in terms of packet loss rate and packet delay when congestion occurs. In wired networks, the Adaptive Randomly Early Detection (ARED) scheme was found to improve TCP fairness. In this paper, we will investigate the effectiveness of ARED scheme when running on individual queues in wireless nodes. We will then propose a Neighborhood Adaptive RED (NARED) scheme, which extends the ARED concept to the distributed neighborhood queue. Simulation studies will be conducted to confirm that the NARED scheme can improve TCP unfairness substantially in ad hoc networks. Moreover, the NARED scheme will be designed at the network level without MAC (Medium Access Control) level protocol modifications. This will considerably simplify its deployment in ad hoc wireless networks. The results will be compared with that of the newly proposed technique namely neighborhood RED which extends the RED algorithm to the distributed neighborhood queue.

Digital watermarking involves the insertion of some information into the electronic contents of image, audio and video data for copyright protection. However, the insertion of such information should not alter the perceived quality of the electronic content (the transparency requirement) while being extremely robust to distortions and malicious attacks (the robust requirement) and while making it impossible to insert another watermark concerning rightful ownership (the maximal capacity requirement). In the last few years the amount of research in digital video watermarking technology for copyright protection has risen significantly due to the demand for such technology by the communication industry that provides multimedia services using digital video coding standards. The video watermarking used in applications like digital television, broadcast monitoring, and streaming video demands more requirements than still image watermarking due to temporal domain attacks and its possible real-time transmission over unreliable and unsecured wired and wireless networks. In this project we will propose a robust digital video watermarking technique with three different properties. Firstly it will embed a maximal capacity invisible watermark without degrading the quality of the video. Secondly it will allow the copyright owner of the video to detect the maximum watermark from the watermarked video that was distorted by video processing operations and network transmission errors and that was altered or removed by the malicious attacks. Thirdly it will provide maximum resistance to the network errors and to the watermark attacks not using the actual parameters of the watermark embedding process. The watermark distortions considered in this project fall into two categories and they are video compression distortions and network transmission distortions. The video compression distortions are caused by the quantization process used in video coding schemes such as MPEG2, and the network transmission distortions are caused by the network errors such as bit error, burst error and packet loss. Similarly the watermark attacks considered in this project fall into four categories and they are desynchronized attack (also called geometric attack), removal and interference attack, cryptographic attack and protocol attack.

The objective computational model for evaluating the visual quality of an image sequence has become an increasingly important in digital video coding and communications. These objective computational models include two classes of objective measures (i) mathematically defined measures such as widely accepted peak signal-to-noise ratio (PSNR) and (ii) human visual system (HVS-based) measures that incorporate perceptual quality measurements. The most recent study by individual researchers and research groups show that the existing HVS-based objective measures do not provide clear advantage over MSE or PSNR. In this project our goal is to develop a simple and computationally inexpensive objective video quality metric that is mathematically defined as well as HVS-based, and provides high correlation with both mathematical and subjective measures. This is an important research problem because such a metric is required to (i) evaluate different video coding and communications systems accurately; (ii) develop better quality metrics for emerging video coding techniques; (iii) monitor broadcast transmissions efficiently and (iv) ensure the most efficient compression of sources and utilization of communication bandwidths.

Details,Going to be added soon