The Foundation: Designing the Heart and Brain of Our System
Every radar system needs a brain – something that can process massive amounts of data in real-time and make sense of it all. As Chip Design Team Lead at Arbe, I design and implement the RPU (radar processing unit) that executes our unique algorithms, which keeps us ahead of the competition. My team delivers a wide range of digital IPs across Arbe’s chipset, whether integrated into our main processor or embedded within our RFICs. Every decision we make directly impacts the performance of the radar system in real vehicles.
Real Problems Require Creative Solutions
Here’s the reality of chip design: every generation brings new performance requirements that push the boundaries of what’s possible. My work involves constantly optimizing architectures to meet demanding frequency, power & area requirements while maintaining system performance. For example, developing the RPU of the Everest 1.1 processor required me to completely redesign the architecture to achieve the frequency targets needed for optimal radar processing. Another major challenge I tackle is defining the “image turn around” process, which is essentially how our processing units access DDR memory when handling the data we sample during radar operation. I designed various access methods that achieved optimal system performance, which required carefully orchestrating how multiple processing units share memory bandwidth without creating bottlenecks.
The Engineering Reality: Making Complex Systems Actually Work
Designing Arbe’s logic IPs is incredibly challenging work that requires our team to constantly suggest creative ideas and solutions for complex problems, throughout the entire development flow – from micro-architecture design all the way till full front-end and back-end integration. This isn’t solo work; developing the various algorithms & system implementations, means collaborating closely with other engineering teams – Research, SW, Analog, System & Hardware. Each group has different priorities and constraints, and my job is making sure our solution meets all requirements & methodologies. The responsibility is significant, and we put serious effort, but seeing our designs work as part of the complete radar system makes it all worthwhile. When you know that the design you developed is actually processing radar data in cars on the road, helping drivers stay safe – that’s what makes the late nights and complex problem-solving worth it.
