Ivica Rimac

A survey on dependable routing in sensor networks, ad hoc networks, and cellular networks

The class of wireless and mobile networks features a dissimilar set of characteristics and constraints compared to traditional fixed networks. The various dimensions of these characteristics/constraints strongly influence the routing system, which is often regarded as the glue of a network. We introduce the concept of routing dependability describing the trustworthiness of a routing system such that reliance can justifiably be placed on the consistency of behavior and performance of the routing service delivered. We investigate this concept by analyzing the basic characteristics of various networks. Subsequently, we derive the most important attributes and impairments that contribute to routing dependability in sensor networks, ad hoc networks, and infrastructure-based cellular networks. Departing from state-of-the-art network designs, we extend our survey to cover future network architectures as well. We finish by briefly investigating possible directions and means that allow mitigating the deprivation of dependability.

Interactive panning and zooming the virtual camera

We explain how camera pans and zooms in digital video can be used to create a panoramic view of a video clip reordering the frames. As the view will normally be too large to be displayed on a single screen we propose the use of a grid where the user can move a virtual camera in different directions which are defined by camera pans as well as by camera zooms. To achieve this goal we explain in detail the architecture of our virtual camera.

Automatic Recognition of Camera Zooms

In this paper we explain how camera zooms in digital video can be recognized automatically. To achieve this goal we propose a new algorithm to detect zooms automatically and present experimental data on its performance.

Observations on Equation-Based Estimation of TCP-Compatible Rate for Multi-rate Multicast Scenarios

In this paper, we investigate the applicability of the equation-based approach to TCP-fair rate estimation for multi-rate multicast rate adaptation and congestion control in principle. The benefit of such an approach is two-fold: on the one hand, participating receivers might base their join and leave decisions on the calculated rate, instead of probing for the appropriate subscription level; on the other hand, the sender side might adapt and optimize the data rates according to receivers’ feedback.

Is dynamic multi-rate multicast worthwhile the effort?

To accommodate heterogeneous transmission conditions in a streaming scenario several multi-rate multicast solutions have been proposed, based on simulcasting or hierarchical layering. At present, most of these schemes follow a receiver-driven layered multicast approach, where the receivers join or leave a subset of the sessions fixed-rate layers in response to changing network conditions. Yet, recently fine-grained coding schemes are being developed, e.g., as proposed in MPEG-4. This will permit a sender to dynamically adapt the size of the layers according to the reported transmission conditions. In this paper, we briefly discuss a general multi-rate congestion control protocol based on dynamic layering and present the basic design challenges. Since adaptation involves the sender and makes dynamic layering fundamentally more complex than its static counterpart, we compare both approaches by extensive simulations in order to explore the theoretical benefit of dynamic over static layering. The main contribution of this paper is to quantitatively describe the performance of both approaches in different scenarios by means of an inter-receiver fairness measure that captures the collective satisfaction of the session receivers.

Music retrieval in ICOR

In this paper we describe music retrieval in ICOR, a project of Darmstadt TU. It is the goal of ICOR to find new interfaces to support applications of music video and music CDs. Although the project consists of audio and video analysis we concentrate on a description of the audio algorithms in this paper. We describe our MPEG-7 like data structure to store meta information for music pieces and explain which algorithms we use to analyze the content of music pieces automatically. We currently use an applause detection to distinguish live music from studio recordings, a genre classifier to distinguish pieces with beats form classical music, and a singer recognition.

Equation-based approach to TCP-compatible multicast congestion control for layered transmission in low-multiplexing environments

Multi-rate multicast has been proposed as a scalable solution to transmitting video over the Internet to receivers with heterogeneous and dynamic rate requirements. The applicability of an equation-based mechanism to congestion control for a protocol which bases its join and leave actions on the calculation of the TCP response function is investigated. We focus on the rate calculation algorithm proposed in TFRC (TCP-friendly rate control), as it is currently a very promising and mature approach to calculating a TCP-compatible rate. By means of a network simulator and an adjusted TFRC protocol implementation, we show that the TCP-compatible rate calculated with the algorithm as originally proposed tends to be biased when applied in environments with a low degree of statistical multiplexing. To improve the performance of the basic algorithm, we propose a simple heuristic approach.

The ICOR framework: a top-down approach to media indexing and retrieval

In this paper, we present the ICOR (Intelligent COntent and Retrieval) framework, which enables the user to easily integrate content description schemes and content analysis algorithms. We show how the generic framework can be adapted to provide appropriate meta-data for applications from different areas, such as music clips or sports games.