Signal Processing and DSP
5G, NB-IoT, LTE, RADAR, Powerline Communications
Azcom Technology experienced software and hardware team has a proven track record in complex Digital Signal Processing (DSP) system design and development. For the last 20 years we have been collaborating with major OEMs and device vendors to design and develop signal processing systems and software across major application domain like wireless, radar, automotive, medical, test and measurement, power line communication.
In wireless domain, Azcom has developed PHY software including algorithms and simulation framework for 5G, NB-IoT, LTE on different DSPs, FPGAs and GPP.
Our Skillset
Azcom offers an outstanding DSP and FPGA firmware development skillset which includes
- Fixed point programming of complex algorithms and bit-exact testing
- Performance/complexity/resources consumption critical evaluation
- State-of-the-art development and testing methods
- Rigorous optimization and refactoring cycles with automated Unit Testing
Complete Life Cycle Management
Simulation Frameworks
We adopt industry standard simulation toolboxes, like MathWorks MATLAB, to cover the entire algorithm design life cycle, starting from the computational complexity and performance evaluation of different alternatives, till the implementation of the best candidate to be the reference for the software unit implementation.
In wireless domain we own complex physical layer link-level simulators (5G, NB-IoT, 4G, HSPA+) with end-to-end simulation capability including UE side, radio propagation, interference conditions.
Platform Expertise – DSPs, Vector Processor, GPP, FPGAs
Azcom has experience in developing signal processing system and software across all the major platforms. These include specialized DSPs from Marvell, Texas Instruments’, ST Microelectronics; Xilinx, Altera and Lattice FPGAs; SoCs built around Tensilica / CEVA DSP cores, ARM NEON architecture processors, Intel Xeon scalable processors.
Signal Processing and Machine Learning
Signal processing is the key to embed Machine Learning on embedded or IoT devices, characterized by limited processing capabilities and power consumption constraints.
DSP can help extracting useful features from raw sensors data, reducing the complexity of the machine learning model (e.g. less hidden layers in DNNs), resulting in improved application latency and decreased connectivity and cloud computing access costs.
Azcom Technology, with its broad DSP skillset on algorithm design and implementation on target hardware, can design and realize the entire DSP pipeline (hardware and software) to evolve legacy rule-based systems (e.g. detection system based only on a threshold level) towards more effective machine learning algorithms.
Research Partnerships
Collaboration with Politecnico di Milano and other universities, like the University of Edinburgh, for specific signal processing research topics (e.g. 5G NR channel coding and decoding, NR receivers design, cellular radio resource management, biometrical signs monitoring).
References
5G PHY, both High and Low PHY
Automotive in-cabin sensing with Radar sensors
Power Line Communication PHY and L2/L3 Stack
LTE PHY
NB-IoT PHY
HSPA+ PHY
