How have the jobs of analog and digital designers changed over the last decade, and what are the main challenges they face today?
The increasing level of integration of analog and digital technologies makes it necessary for designers in each domain to understand and account for the other. High-speed mixed-signal electronics, digitally-controlled RF transceivers, hybrid beamforming for 5G massive MIMO antenna arrays and digital linearization of RF power amplifiers are just a few examples. In each of these cases, it is impossible to ignore RF effects in the digital design and vice versa.
For these types of designs, engineers need to work at a higher level of abstraction, to model both the analog and digital components together and to simulate fast enough to get results for complex multi-domain designs. In many companies, this is causing a re-evaluation of traditional workflows that design digital and analog components separately.
How has MathWorks’ software evolved over the last few years to meet the changing needs of designers?
MathWorks’ software has evolved in multiple directions to meet these changing needs. Some recent enhancements include:
- Products for simulating digital, RF and antenna components together, allowing earlier optimization and verification of complex designs.
- LTE and WLAN standards-compliant simulation and waveform generation for design and verification, including over-the-air testing using software-defined radios and RF instruments.
- Optimized HDL generation, from models and a hardware/software co-design workflow for prototyping and implementation.
- Parallel computing solutions for large-scale simulation, from desktop to cloud services.
Can you give us a little background on MATLAB and Simulink applied to wireless applications and how they work together?
MATLAB is widely used by wireless engineers and researchers as a programming language for rapid algorithm development, analysis and visualization. The MATLAB language and toolboxes provide a large set of functions and apps for wireless engineering tasks.
Wireless engineers use Simulink to model and efficiently simulate systems that incorporate digital and RF components, adaptive systems with feedback control and bit-accurate and timing-accurate hardware design. Simulink is the foundation for model-based design, which uses system models to generate and verify hardware and embedded software.
Together, MATLAB and Simulink combine textual and graphical programming to design systems in a unified simulation environment. MATLAB code can be used to create Simulink blocks, generate input data to drive simulations, run simulations in parallel and analyze and visualize simulation results.
How does MathWorks help analog and digital design teams work together and improve efficiency?
Analog engineers can create and share models for multidomain simulation of hybrid analog and digital designs. These common models help the engineers characterize the impact of their designs on adjacent components and overall system performance.
These models operate at a higher level of abstraction than RTL-level or circuit-level simulations. As a result, simulations run many times faster to allow investigation of many more scenarios. Problems can be found and corrected before going to the test lab or building hardware.
What are some of the challenges facing designers doing 5G systems?
The primary goal of 5G new radio (NR) technology in 3GPP release 15 is to increase the data rate and capacity of mobile broadband networks, which will be achieved by moving to higher frequencies and larger bandwidths.
The release 15 standard introduces several changes to the physical layer to accomplish these goals, including a more flexible numerology and frame structure and new channel coding schemes. These changes add complexity to the design that will require different architectures to contain cost and achieve performance goals.
Another challenge will be achieving power efficiency and linear performance across wide bandwidths in the RF front-end, including power amplifiers. This requires adaptive DSP techniques such as digital predistortion (DPD), which often must be designed and verified in simulation before the RF hardware is available.
Perhaps the biggest challenge is the design of massive MIMO antenna arrays and beamforming techniques to overcome high frequency propagation losses. This problem brings together multiple engineering disciplines: signal processing, RF and antenna design. It requires design tools that enable modeling and simulation across these domains to select the right beamforming architecture (digital, analog or hybrid) that will optimize cost and performance for the desired operating frequency.
Finally, designers need to characterize RF signal propagation channels in various outdoor and indoor scenarios. The physics of RF propagation at 5G millimeter wave frequencies required the development of new channel models that are essential to ensure that designs can operate in realistic conditions. These models enable this design verification step to be done with simulation, rather than waiting for expensive and time-consuming lab and field tests.
With phased array technology being commercialized today, are there lessons learned from the defense industry?
- Beamforming is the most common and most effective technique for achieving high signal gain when phased array antennas are used.
- More benefits of phased array technology can be realized when it is designed and implemented as an integrated part of the whole system.
- Digital technology has made more advanced phased array techniques plausible. All-digital techniques provide greater accuracy (e.g., beam steering in both azimuth and elevation).
- Hybrid implementation (analog and digital together) of phased array technology is often more cost effective and should be considered for large arrays and millimeter wave frequencies.
- The evaluation of different digital and hybrid architectures can be done more efficiently by simulating the digital and RF components together.
What are some of the challenges ahead in vehicle advanced safety systems and autonomy?
The list is long. Automated driving spans a wide range of automation levels — from advanced driver assistance systems (ADAS) to fully autonomous driving. These advanced safety and automated driving systems use vision, radar, LiDAR and combinations of sensor technologies (and soon, vehicle-to-vehicle communications) to automate dynamic driving tasks. These tasks include steering, braking and acceleration.
Design of the component technologies in automated driving systems can only be validated in the context of integrated simulations that include:
- Vision, radar and LiDAR perception algorithms.
- Sensor fusion and controls development.
- Wireless communication (V2V, V2X) using direct or cellular-based protocols.
- Ground-truth labeling and re-simulation.
- Deep learning algorithms for object recognition.
- Environment, vehicle and driver models.
- Conformance to ISO 26262 requirements.
How Does MathWorks help transition the models and simulations to testing and production?
MathWorks offers software that helps engineers implement model-based design of wireless systems.
With model-based design, system architects and hardware engineers can use MATLAB and Simulink models for each of their tasks and share them as a design progresses from abstract models to production. Hardware-accurate Simulink models can automatically generate readable, synthesizable HDL code for FPGA, SoC and ASIC implementation. System architects can build prototypes with popular FPGA and software-defined radio kits and hardware engineers can reuse those models for production deployment.
HDL-optimized LTE IP blocks and verified reference applications help improve LTE system performance and shorten development time.
MATLAB and Simulink help to automate testing to verify that designs function correctly, before hardware implementation. Engineers can use the validated models as a test bench for verifying hardware prototypes and production implementations. They can test designs using a range of SDR hardware and RF instruments from vendors of their choice, automatically generate SystemVerilog models for ASIC verification and efficiently analyze large datasets from simulations, lab tests and field trials.
MathWorks is one of the few companies that supply an algorithm to antenna simulation solution. How does that benefit the designer?
Algorithm-to-antenna simulation is accomplished using multidomain models of the highly integrated technologies being developed for 5G. Baseband, RF and antenna engineers can use these simulations to help them design technologies such as massive MIMO arrays, hybrid beamforming architectures and adaptive RF transceivers and front-ends, all in the same environment.
Wireless designers can employ high-level and high-fidelity models to realistically simulate component interactions, quickly evaluate design tradeoffs and analyze the performance impact of design choices. By testing with multidomain simulation, they can find errors sooner, spend less time debugging in the hardware lab and respond faster to new requirements.