Viktoria Fodor

Resilience in live peer-to-peer streaming [Peer-to-Peer Multimedia Streaming]

The success of peer-to-peer overlays for live multicast streaming depends on their ability to maintain low delays and a low ratio of information loss end-to-end. However, data distribution over an overlay consisting of unreliable peers is inherently subject to disturbances. Resilience is thus inevitably a key requirement for peer-to-peer live-streaming architectures. In this article, we present a survey of the media distribution methods, overlay structures, and error-control solutions proposed for peer-to-peer live streaming. We discuss the trade off between resilience and overhead and argue that efficient architectures can be defined only through thorough performance analysis.

On the Performance of Multiple-Tree-Based Peer-to-Peer Live Streaming

In this paper we propose and analyze a generalized multiple-tree-based overlay architecture for peer-to-peer live streaming that employs multipath transmission and forward error correction. We give mathematical models to describe the stability properties of the overlay and evaluate the error recovery in the presence of node dynamics and packet losses. We show how the stability of the overlay improves with the proper allocation of the outgoing bandwidths of the peers among the trees without compromising its error correcting capability.

Detecting Low-Power Primary Signals via Distributed Sensing to Support Opportunistic Spectrum Access

Cognitive radio operation with opportunistic spectrum access has been proposed to utilize spectrum holes left unused by a primary system owning the spectrum license. The key of cognitive radio operation is the ability to detect weak primary signals and to control the transmission of cognitive users in a way that interference between the two systems is minimized. In this paper we evaluate how a sensor network deployed to provide distributed spectrum sensing can assist cognitive operation. Specifically, we consider sensor networks with regular topology, where a high level of cooperation also means that sensors far from the source of the primary signal are involved in the sensing process. Assuming energy detection and hard-decision combining we derive worst case probabilities of missed detection and false alarm, determine the necessary level of cooperation among the sensors and evaluate how the sensor density and the sensing time affect the performance of distributed sensing.

Probe-based admission control for multicast

End-to-end measurement based admission controls (MBAC) have recently been proposed to support quality of service for real-time transfer of data. All these designs share the idea of decentralizing the admission decision by requiring each end host to probe the network before transmission. These schemes are solely targeted at unicast communications, while multicast data has not yet been addressed. We study a probing procedure to perform admission decisions for multicast senders and receivers. The admission control offers a reliable upper bound on the packet loss for the multicast session even with short probe phase durations (e.g. half a second). Our probing mechanism only requires the routers to differentiate between two classes of packets: high priority data and low priority probes. Simulation results of the performance of the procedure are presented and evaluated.

On the effects of the packet size distribution on FEC performance

For multimedia traffic like VBR video, knowledge of the average loss probability is not sufficient to determine the impact of loss on the perceived visual quality and on the possible ways of improving it, for example by forward error correction (FEC) and error concealment. In this paper we investigate how the packet size distribution affects the packet loss process, i.e., the probability of consecutive losses and the distribution of the number of packets lost in a block of packets and the related FEC performance. We present an exact mathematical model for the loss process of an MMPP -- MMPP/ Er/1/K queue and compare the results of the model to simulations performed with various other packet size distributions (PSDs), among others, the measured PSD from an Internet backbone. The results show that analytical models of the PSD matching the first three moments (mean, variance and skewness) of the empirical PSD can be used to evaluate the performance of FEC in real networks. We conclude that the exponential PSD, though it is not a worst case scenario, is a good approximation for the PSD of todays Internet to evaluate FEC performance. We also conclude that the packet size distribution affects the packet loss process and thus the efficiency of FEC mainly in access networks where a single multimedia stream might affect the multiplexing behavior. We evaluate how the PSD affects the accuracy of the widely used Gilbert model to calculate FEC performance and conclude that the Gilbert model can capture loss correlations better if the CoV of the PSD is high.

Real-Time Distributed Visual Feature Extraction from Video in Sensor Networks

Enabling visual sensor networks to perform visual analysis tasks in real-time is challenging due to the computational complexity of detecting and extracting visual features. A promising approach to address this challenge is to distribute the detection and the extraction of local features among the sensor nodes, in which case the time to complete the visual analysis of an image is a function of the number of features found and of the distribution of the features in the image. In this paper we formulate the minimization of the time needed to complete the distributed visual analysis for a video sequence subject to a mean average precision requirement as a stochastic optimization problem. We propose a solution based on two composite predictors that reconstruct randomly missing data, and use a quantile-based linear approximation of the feature distribution and time series analysis methods. The composite predictors allow us to compute an approximate optimal solution through linear programming. We use two surveillance videos to evaluate the proposed algorithms, and show that prediction is essential for controlling the completion time. The results show that the last value predictor together with regular quantile-based distribution approximation provide a low complexity solution with very good performance.

Characterization of SURF interest point distribution for visual processing in sensor networks

We study the statistical characteristics of SURF interest points and descriptors, with the aim of supporting the design of distributed processing across sensor nodes in a resource constrained visual sensor network. We consider a sensor network with a single camera node and four schemes of delegating processing tasks to the sensor nodes. We discuss the potential and the challenges of the different schemes in light of the results of the statistical analysis. Our results show that the distribution of the number of interest points per image exhibits a heavy tail. The interest point locations are almost uniformly distributed along the axes of the images, but their X and Y coordinates are slightly correlated. Most interest points are found in the lowest octave layers, and the number of interest points decreases exponentially with scale. Our analysis suggests that for a wireless broadcast channel delegating subareas of images to processing nodes would lead to a more even allocation than delegating by octave layers. For directional wireless channels the efficiency can be significantly improved by performing some of the feature extraction tasks at the camera node.

Cooperative image analysis in visual sensor networks

This work addresses the problem of enabling resource-constrained sensor nodes to perform visual analysis tasks. The focus is on visual analysis tasks that require the extraction of local visual features, which form a succinct and distinctive representation of the visual content of still images or videos. The extracted features are then matched against a feature data set to support applications such as object recognition, face recognition and image retrieval. Motivated by the fact that the processing burden imposed by common algorithms for feature extraction may be prohibitive for a single, resource-constrained sensor node, this paper proposes cooperative schemes to minimize the processing time of the feature extraction algorithms by offloading the visual processing task to neighboring sensor nodes. The optimal offloading strategy is formally characterized under different networking and communication paradigms. The performance of the proposed offloading schemes is evaluated using simulations and is validated through experiments carried out on a real wireless sensor network testbed. The results show that the proposed offloading schemes allow to reduce the feature extraction time up to a factor of 3 in the reference scenario.

On the stability of end-point-based multimedia streaming

In this paper we propose an analytical model of a resilient, tree-based end-node multicast streaming architecture that employs path diversity and forward error correction for improved resilience to node churns and packet losses. Using the model and via simulations we study the performance of this architecture in the presence of packet losses and dynamic node behavior. We show that the overlay can distribute data to nodes arbitrarily far away from the root of the trees as long as the loss probability is lower than a certain threshold, but the probability of packet reception suddenly drops to zero once this threshold is exceeded. The value of the threshold depends on the ratio of redundancy and on the number of the distribution trees. Using the model and simulations we show that correlated and inhomogeneous losses slightly worsen the overlay’s performance. We apply the model to study the effects of dynamic node behavior and compare its results to simulations.

Prediction-based load control and balancing for feature extraction in visual sensor networks

We consider controlling and balancing the processing load in a visual sensor network (VSN) used for detecting local features, such as BRISK. We formulate a prediction problem with random missing data, and propose two regression-based algorithms for data reconstruction. Numerical results illustrate the performance of the proposed algorithms, and show that backward regression combined with the last value predictor can be used for controlling and balancing the processing load in VSNs with good performance.