MulticoreWare

Case Studies

DSP Optimization of RADAR Perception Software

September 14, 2022

The client is a US-based leading technology company that develops sensors, sensor-based solutions, sensor software, and other mission-critical products. The client has shipped greater than one million sensor units in about 10+ countries in the year 2021. The client was looking for a technology partner to implement software acceleration for RADAR based software on DSPs.

The Project

Our customer’s R&D team had dedicated efforts to commercialize their next generation RADAR sensor solution. The customer had developed sophisticated home-grown perception software to be executed on the radar raw data to identify objects and tracks in an automotive environment. The custom designed algorithm was developed on a CPU platform and used compute-intensive mathematical operations. The requirement of the customer was to optimize this algorithm and reduce the latency for a specific low-powered target DSP.

Constraints:

The algorithm was home-grown and expected to be continually updated during the course of the project to improve accuracy performance.

The runtime throughput expectations were very aggressive to satisfy the safety needs of the automotive industry.

The original software design was compute and memory intensive.

The MulticoreWare Advantage & Approach

Multicoreware had the unique edge of having a strong RADAR perception team of specialists in addition to decades of experience in developing and customizing software for power-constrained compute environments. In order to speed up the perception software, Multicoreware collaborated extensively with the customer to understand the design of the core algorithms and to recommend areas for enhancement and re-implementation.

MCW Approach

  • MCW conducted technical study of the target platform and the DSP intrinsics to identify the possibilities of optimization from a hardware perspective. 
  • MCW reported the most memory intensive blocks of the algorithm software design and reimplemented to achieve maximum memory efficiency. 
  • MCW established a cross-geographical team of specialist engineers and solutions architects from USA, China and India and executed the project. 
  • MCW exceeded the client’s expectations by achieving 3X more speed-up than initial requirements. 

OUTCOME

Delivered a fully functional, DSP optimized RADAR perception software that is to be integrated with our client’s sensor solution and shipped to the automotive industry. 

Having a RADAR – Perception Software Problem? Talk to our specialists.

Share Via

Explore More

May 8 2026

Optimizing Android Application Performance for Remote GPU Rendering Platforms

Customer
The customer is a technology company specializing in GPU virtualization middleware that enables discrete processing units to be aggregated into shared resource pools and accessed remotely across conventional network infrastructure.

Read more
Apr 9 2026

Agentic AI for RAN Observability, Explainability and Orchestration

Customer A global telecommunications and network infrastructure company that provides advanced software, hardware, and services for building, managing, and optimizing large-scale telecom and enterprise networks. Its solutions leverage AI, automation, and end-to-end visibility to help operators enhance performance, ensure reliability, and efficiently manage complex, multi-domain network environments. Problem Statement Radio Access Networks (RAN) are the  … Read more

Read more
Apr 3 2026

Embedded Platform Optimization for Advanced Drone Systems: Lidar and Motor Control Integration

Client A leading drone and robotics company developing high-performance UAV platforms for autonomous operations, industrial inspection, and surveying in complex or restricted environments. Problem Statement Simultaneously executing high-throughput LiDAR processing and latency-critical motor control on resource-constrained embedded systems creates a fundamental bottleneck in real-time performance and scalable UAV autonomy. Challenge 1: High-Speed Sensor Integration Integrating  … Read more

Read more

GET IN TOUCH

    Please note: Personal emails like Gmail, Hotmail, etc. are not accepted
    (Max 2000 characters)