LOFAR, the LOw-Frequency ARray, is a new-generation radio interferometer constructed in the north of the Netherlands and across europe. Utilizing a novel phased-array design, LOFAR covers the largely unexplored low-frequency range from 10-240 MHz and provides a number of unique observing capabilities. Spreading out from a core located near the village of Exloo in the northeast of the Netherlands, a total of 40 LOFAR stations are nearing completion. A further five stations have been deployed throughout Germany, and one station has been built in each of France, Sweden, and the UK. Digital beam-forming techniques make the LOFAR system agile and allow for rapid repointing of the telescope as well as the potential for multiple simultaneous observations. With its dense core array and long interferometric baselines, LOFAR achieves unparalleled sensitivity and angular resolution in the low-frequency radio regime. The LOFAR facilities are jointly operated by the International LOFAR Telescope (ILT) foundation, as an observatory open to the global astronomical community. LOFAR is one of the first radio observatories to feature automated processing pipelines to deliver fully calibrated science products to its user community. LOFAR's new capabilities, techniques and modus operandi make it an important pathfinder for the Square Kilometre Array (SKA). We give an overview of the LOFAR instrument, its major hardware and software components, and the core science objectives that have driven its design. In addition, we present a selection of new results from the commissioning phase of this new radio observatory.

No abstract

The performance of a high throughput and large-scale signal processing system must not be compromised by the control and monitoring flow that is inherently part of the system. In particular, the interfacing of data flow and control flow components should be such that control does not obstruct the signal flow that is of higher priority. We assume that the signal processing is modeled as a distributed hierarchy of data flow networks, and that the control and monitoring is modeled as a distributed hierarchy of communicating Finite State Machines. The interfaces between leaf-nodes of the control and monitoring network, and the signal processing nodes in the dataflow networks are specified in such a way that the semantics of both network types are preserved. In this paper, we present the prototyping of a control network and its interfacing with a data flow network in a FPGA-based platform, and we analyze the performance of the interfacing in a case study. The HDL code that is involved in the interfaces is generated in a semi-automated way.

No abstract

When mapping large and high-throughput signal processing applications onto heterogeneous platforms, parts of these applications are assigned to re-configurable components. Automating such mappings without delving deep into details implies the (re-) use of IP components. When it comes to 1) relying on IP components in systemlevel and (re-configurable) platform-based design, and 2) porting of such designs across platforms, it is not well known how to integrate both IP libraries and portability requirements into the design flow. To investigate these uncertainties, we have conducted four case studies around the (re-)use and interfacing of IP components. One is focusing on the porting issue, one is dealing with a new automated task-level mapping method, one is evaluating a HW-SW commercially available co-design method, and one is about standardization of interfaces for IP wrapping. The case studies reveal the weakness of otherwise highly desirable system-level design methods when evaluated with respect to fast, accurate, and systematic IP integration.