Digital signal processing
Digital signal processing (DSP) has become an essential part of nearly every product in today’s world. Any device that contains wireless connectivity, audio, video, measurement sensors or automatic control will require some form of DSP.
Our team of experts combine extensive knowledge of digital signal processing, complex software development and state-of-the-art processors with established tools and processes for developing ground-breaking and highly differentiated products for our clients.
Software-defined radio (SDR)
SDR has revolutionised communication systems in recent years – dramatically reducing timescales and cost, and eroding the dominance of digital ASICs.
Many of today’s standards-based communications systems are continually evolving and therefore benefit greatly from the flexibility of SDR, allowing enhancements and bug fixes to be incorporated with relative simplicity. This is particularly apparent in cellular (2G, 3G and 4G) standards and satellite communications.
Advanced DSP architectures
The early digital signal processors have now evolved into highly complex system-on-chip (SoC) integrated circuits featuring multi-core architecture and acceleration engines to offload processing to more dedicated hardware units.
To achieve real-time operation while optimising the use of processing resources, our development teams carefully architect the firmware to complement the hardware. Understanding how DSP algorithms can be efficiently implemented within cycle and memory constraints and incorporating diagnostic facilities into the code are essential to enabling detailed visualisation of program execution.
Simulation & modelling
We research and design algorithms using the industry standard tools provided in MathWorks MATLAB. Through rapid development, analysis and iteration of the algorithms, our development teams can quickly see if a design is viable before proceeding with implementation, which usually bears the biggest risks and challenges.
The engineers involved in the design of these algorithms are also deeply knowledgeable in the processing and memory constraints of hardware, thereby minimising the risk that designs cannot be implemented.
Digital audio and digital communications frequently combine within the products we develop – for example, in the audio codecs which carry voice over a satellite link or high-quality music over a short-range Bluetooth link.
Frequently, signal conditioning (such as acoustic echo cancellation or beamforming) may be required to enhance the audio and improve signal quality. Many of the applications we address involve significant power constraints, adding another dimension to algorithm and processing hardware design.
Track record of success
Over the last 30 years, our DSP expertise has been at the heart of many breakthrough projects, from the Prism 200 through-wall radar to the Videocore low-power multimedia processor that we spun out to form Alphamosaic Ltd.
These successes were made possible by the unique combination of skills we possess, bringing together expertise in mathematics and algorithm design, complex software development and software/hardware integration.
Pioneering the future
We remain at the forefront of DSP technology, pioneering the use of highly integrated, application-specific programmable processors, enabling the adoption of software-defined radio in wireless, cellular and satellite networks, and using small, low-power MEMS sensor in innovative location, tracking and medical applications.
Our considerable DSP skills are now being applied to problems that incorporate cutting-edge computer vision, machine learning and AI technologies.
Tech, tools & facilities
Scalable software test facility
Continuous integration (CI) is central to building and delivering high-quality software. Our virtualised CI test facility – codenamed Smokey – enables us to test complex software at scale, with or without hardware in the loop, to identify problems earlier and fix them faster. Smokey provides a clean, isolated virtual build environment, pre-loaded with industry-leading tools such as Coverity static analysis. CI testing is integrated with our version control, enabling every software revision to be automatically built, tested on real hardware and analysed.
Development & test laboratories
We have 100,000sq feet of laboratories spread across the UK and USA, including electronic production, fabrication and low-volume production capability. This includes medical laboratories specifically equipped for analytical testing of drug delivery devices, from prototype development to design verification. Our climate-controlled labs feature laser diffraction, high-speed video, in-vitro test equipment, metrology, microscopy and force test equipment.
Design & analysis tools
We use a range of design tools including Solidworks and Pro/E. Analysis tools include Ansys Fluent for fluid analysis, Ansys Mechanical for finite element analysis and Maxwell for electromagnetic analysis. We use Zemax for optical simulation, Matlab for general mathematical/statistical simulation and Labview for the rapid build and control of laboratory equipment.
The Digital Greenhouse is our purpose-built AI research facility, specifically designed to discover, develop and test machine learning approaches in a fast, secure environment. The facility features an NVIDIA DGX-1 Deep Learning Supercomputer, a bank of machine learning computers, petabyte local storage and many teraflops of dedicated compute power.
Our flexible, responsive and agile manufacturing facilities are ideally suited to electronic and electromechanical product builds, enabling us to very quickly set up short product runs built to an extremely high quality. We can build our own PCBAs using industrial representative equipment and ensure manufacture-ability during design. Having our own EPD facility is also extremely helpful in development, driving out DfX activities in eCAD, and feeding test results back into the design process from an early stage.
We use a range of specialist software for model-based design and simulation. These include MathWorks Matlab and Simulink (including automated code generation and GPU acceleration), dSPACE for model based design, hardware in the loop testing and development of mechatronic systems, and CFD solutions including COMSOL Multiphysics for multiphase flow simulation. Finite element packages like ANSYS Maxwell and HFSS, CST Studio and Microwave Office are used for full-wave, transmission line and magnetostatic simulation. We also employ a dedicated compute cluster for machine learning, neural network training and continuous verification.
We can help you achieve the impossible