Carsten Grenz

Distributed three-dimensional camera alignment in highly-dynamical prioritized observation areas

Video surveillance of large areas has become a necessary safety procedure. With an increase of these areas and an accompanying increasing number of cameras, manual configuration becomes infeasible. This paper describes a way to automate this task by distributing it among the smart camera nodes. The main objective is an overlap-free monitoring of the observation area, considering the distinct priority of an area element. Due to a time-dependable priority the observation quality will increase in contrast to static priorities. The cameras only use local knowledge and single-hop communication. Each camera optimizes its local field-of-view according to a specified quality function. It is shown that the developed algorithm scales well to large smart-camera systems.

Towards organic active vision systems for visual surveillance

Vision systems are camera networks, which perform computer vision using multiple cameras. Due to the directional characteristic of the image sensor, it is not feasible to observe large surveillance areas using stationary cameras. This drawback can be overcome by using active vision (AcVi) systems consisting of mobile cameras (called AcVi nodes), which are reconfigurable in both position and orientation. In case of a dynamic environment with moving targets, AcVi nodes can be repositioned during runtime in order to fulfill the overall observation objectives. This paper is devoted to develop a generalized system architecture for AcVi systems in multi-target environments. Based on this architecture, we present a coordination mechanism making way for self-organization - a major paradigm of Organic Computing - in such systems. Our algorithm enables AcVi nodes to observe an area under surveillance by using their mobility. In contrast to stationary vision systems, only a fraction of the amount of nodes is needed to achieve the same observation quality. The algorithm has been evaluated by simulation with up to 10 nodes as used for example for surveillance scenarios. Results show that our coordination algorithm is able to cope with large numbers of nodes and targets and is robust towards real world disturbances like communication failure.

PhD forum: Adaptive storage management in highly heterogeneous Smart Sensor systems

Future surveillance systems will be highly heterogeneous systems consisting of Smart Cameras mixed with many other types of Smart Sensors. Although the ongoing advances in computer architecture make way for performance improvements even in the smallest sensor nodes, their processing and communication capabilities will still differ from the computation power of Smart Cameras. Moreover, energy consumption is still a major concern of sensor networks. Even with highly developed batteries, reducing the energy consumption wherever possible remains a major task of system architects. Thus, it is an important design criteria for storage control algorithms. This work proposes an adaptive management system which optimizes the storage behaviour of heterogeneous sensor systems while preserving easiest accessibility of the data by the user.

Access-Centric In-Network Storage Optimization in Distributed Sensing Networks

Distributed sensing networks are getting increasingly complex these days. The main reason are the changing demands of the users and application scenarios, which require multipurpose systems. Enabled by continuously improving computational and storage capacities of sensors, this development leads to an increasing number of different algorithms which run concurrently in a sensing network. Thereby, they enable sensor-actuator platforms to perform various kinds of analysis and actions in parallel. Within such a sensor network a variety of algorithms is performed simultaneously. When developing distributed vision and control algorithms, developers focus mainly on the consecutive processing stages. Such a process typically begins with perceiving raw sensor data and terminates with delivering high-level event data to responsible entities. Thereby, different stages may be performed at varying locations within the underlying network. Although the researchers may apply custom optimizations to their data flows, these are highly specific. During design time, it is impossible to anticipate each system environment or predict their algorithms’ possible interactions and synergies with other data flows. We propose a generic storage architecture which separates algorithms from data storage and retrieval. By making use of the fact that most data in sensing networks refers to geographic areas, our architecture takes care of the data flow and its online optimization throughout the network at runtime. By decoupling the processing stages from the data flow, we allow for self-organizing meta-level optimizations of data placement in the network. Moreover, this approach even makes inter-algorithmic optimizations possible, if different algorithms process similar data within their step-wise processing logic. With the introduction of the access-centric storage paradigm, we prove to reduce network load and query latency at the same time at runtime.

Self-Organizing Access-Centric Storage Optimization in Smart Sensor Networks

Sensor networks are getting much more complex these days. The mixture of various low-cost sensors together with increasing computational power enables for whole new systems running a lot of different analysis and control algorithms concurrently. It is impossible to anticipate their composition and data flows a priori. Although the actual data flows are hardly predictable during design-time, we present a lightweight and self-organizing approach on how shared data stores are used to optimize the storage allocation of data during run-time. While mostly using the existing traffic to disseminate routing information, we show that our distributed algorithm significantly reduces query latencies by placing data according to the access-centric storage paradigm.

Task Execution in Distributed Smart Systems

This paper presents a holistic approach to execute tasks in distributed smart systems. This is shown by the example of monitoring tasks in smart camera networks. The proposed approach is general and thus not limited to a specific scenario. A job-resource model is introduced to describe the smart system and the tasks, with as much order as necessary and as few rules as possible. Based on that model, a local algorithm is presented, which is developed to achieve optimization transparency. This means that the optimization on system-wide criteria will not be visible to the participants. To a task, the system-wide optimization is a virtual local single-step optimization. The algorithm is based on proactive quotation broadcasting to the local neighborhood. Additionally, it allows the parallel execution of tasks on resources and includes the optimization of multiple-task-to-resource assignments.

Selbstorganisierende Smart-Kamera-Systeme

ZusammenfassungZukünftige Systeme zur Überwachung von großen Flächen werden auf der Basis von verteilten intelligenten Kamerasystemen entwickelt werden. Aufgaben solcher Systeme sind beispielsweise das Verfolgen und Zählen von bewegten Objekten und die Analyse ihres Bewegungsverhaltens. Jede dieser Smart-Kameras ist ein autonomer Knoten, ausgestattet mit einem Schwenk-/Neige-/Zoom-Aktuator (PTZ für pan/tilt/zoom), Verarbeitungsressourcen und einer Kommunikationsschnittstelle. Dieser Beitrag gibt einen Überblick über das Forschungsgebiet der Smart-Kamera-Systeme und Beispiele für verteilte Steuerungsalgorithmen, welche die systemweite Selbstorganisation ermöglichen. Der Begriff Selbstorganisation beinhaltet einen integrierten Ansatz zur Selbstkonfiguration, Selbstoptimierung (Smart-Kameras konfigurieren und optimieren ihre Sichtbereiche) und Selbstheilung (Smart-Kameras übernehmen Aufgaben ausgefallener Knoten). Smart-Kamera-Systeme werden dabei als verteilte Systeme auf der Basis von Ad-hoc-Netzen modelliert. Diese Architektur erlaubt es, die Nachteile bisheriger zentraler Ansätze in den Bereichen Skalierbarkeit und Fehlertoleranz zu vermeiden.

ENRA: Event-based network reconfiguration algorithm for Active Camera Networks

Smart cameras play an important role for security systems as they combine video sensing, video processing and communication within a single device. One weak point is that smart cameras are usually stationary and so occlusions may create blind spots in the system. These blind spots may be overcome by enhancing smart cameras with mobility (e.g., mounting them on unmanned vehicles) in order to create so called Active Cameras (ACs) and network them into AC networks. Nevertheless, mobility of cameras comes along with challenges in terms of coordination and configuration. Additionally, the per unit cost of AC networks is higher in comparison to static camera networks but the number of cameras needed to observe large areas, e.g. an apron of an airport, can be reduced considerably. In this paper, we present an algorithm paving the way for self-adaptation in AC networks. Based on the number of events occurring in the workspace ACs increase or decrease so-called spatial redundancy regions with neighboring nodes to balance the networks load. Simulation results show that the presented protocol increases the overall systems performance and is well-suited for dynamic scenarios.