COSMOS: Co-existence Simulation Modelling of radars for Self-driving

COSMOS aims to develop an experimentally validated holistic sensing modelling environment, able to represent the radar sensing element of autonomous vehicles with sufficient realism to augment and extend to the full-scale test-track field trials.

Overhead street view of a car with sensors fittingThere is an external sensor level modelling, where each part: sensors configuration, propagation channels, interference effects, signal processing, and performance analysis are modelled together with their interaction to provide a verifiable performance metric and facilitate the development of mitigation strategies.  The project duration is 3 years, with a total budget of £2.9m.



Autonomous vehicles are being developed with the promise of improving road traffic safety, passengers’ conveniences as well as economic and environmental benefits. Autonomous functions that take control of the driving tasks off the human driver are heavily dependent on the system ability to perceive and understand the vehicle's surroundings through complex sensor systems. The sensor suite consists of radar, lidar, and video for situational awareness of the surroundings.

In the heart of autonomy, alongside artificial intelligence are these sensors which not only replace human ability to “see the road” but shall outperform human ability. Most of the sensors are actively emitting signals to use their reflections from the environment to understand the scene. The sensor data-processing algorithms produce understanding of the environment, enabling the autonomous function control system to adapt in response to prevailing conditions.

As connected and autonomous vehicles (CAVs) shall operate 24/7 in any weather and lightening conditions, the most robust and reliable sensor is radar. According to current vision each CAV will be equipped by at least a dozen of radars overlooking 360 degrees surrounding and seeing from a meter to hundreds of metres ahead and around.

  • Co-existence and reliable performance of radars in CAVs are of critical importance for enabling reliable real-world L4 and beyond autonomy, in any environment.
  • Multiple radars gives rise to co-existence challenges (mutual interference,  environmental effects such as multipath, multi-bounce, etc.).
  • Validated simulations are crucial to evaluate autonomous systems performance in complex scenarios and develop safety standards.


  • Improving road safety through the accelerated development of CAV technologies
  • Supporting CAV sensors and algorithms design
  • Providing effective testing and enhanced verification and validation of sensors
  • Mutual interference suppression with cognitive sensors to modify parameters in real time
  • Developing and validating optimal L4/L5 functions and generating recommendations to the wider CAV community



Microwave integrated systems laboratory (MISL). Based within the Department of Electronic, Electrical and Systems Engineering at the University of Birmingham, MISL is the UK's leading academic research group in radar and remote sensing.


•    Jaguar Land Rover (JLR) – industrial lead
•    Igence Radar


The project is collaborative, funded by Innovate UK (Grant number 104526). Innovate UK aims to connect and fund business and research collaborations to accelerate innovation and drive business investment into research and development.


  • Developing simulation environments (system-of-systems approach) to create the whole sensing loop where sensors, control system, signal processing, and environment will interact to emulate driving scenarios.
  • Novel fully customisable holistic simulator for sensor-to-sensor co-existence modelling, analysis, and performance evaluation of millimetre wave automotive radars for L4/L5 autonomy addressing:
    • Physics-based radar sensor scene response for a variety of waveform and radar processing chain parameters, roadside conditions, automotive interferers
    • Complex environment and propagation channels modelling
    • Capability of analysing and post-processing radar data collected in real-time, according to the hardware capabilities of current state of the art automotive sensors
  • Quantitative and qualitative analysis of radar performance and mitigation of radar interferences for robust sensor performance in a co-existent automotive environment

Flow chart showing automotive sensor processes

Caption: University of Birmingham Automotive Radar Simulator Block Diagram

Approach and achievements

Holistic Simulator for sensor-to-sensor co-existence modelling

Modular holistic simulation platform addressing:

  • High fidelity CAV radar modelling
  • Complex environment and propagation channels modelling
  • Models of highway scenarios
  • Multiple radars co-existence modelling & interference mitigation strategies
  • Autonomous system data processing and vehicle control loop
  • Modular open architecture for higher level integration

Caption: University of Birmingham Multi-Dimensional Processing of Automotive Radar Data

Universal Tools for Radar Performance Analysis

  • Novel generalised universal approach to evaluate radar performance in the presence of interference with various waveform parameters
  • Identification of safe regions of operation in co-existence of automotive radars

Two graphs illustrating universal tools for radar performance analysis

Performance evaluation of MIMO and phased array-based radars

  • Novel comparison of state-of-the-art automotive sensors with various antenna configurations operating in a coexisting environment.

Series of instrument measurements showing evaluation of MIMO and phased array-based radars


MIMO Array: 3*4 (Transmit*Receive elements)
Phased Array: 1*12 (Transmit*Receive elements)

Radar Response Modelling in a tunnel

  • Evaluation of extend of ACC saturation in an enclosed structure such as a tunnel

Sequence of diagrams illustrating radar response modelling in a tunnel

Additional project outcomes

Data Repositories

COSMOS dataset  for co-existence/ interference analysis and simultaneous scene representation by automotive radar and video with GPS/IMU ground truth

Extensive field trials performed at the University of Birmingham and HORIBA MIRA to collect the radar data in different road scenariosSelection of images showing field trials assessing automotive radar

Multi-sensor raw data

COSMOS dataset for co-existence/interference analysis and simultaneous scene representation by automotive radar and video with GPS/IMU ground truth:

Processed radar data


  1. Norouzian, F., Pirkani, A., Hoare, E., Cherniakov, M., & Gashinova, M. (2021). Phenomenology of automotive radar interference. IET Radar, Sonar & Navigation.
  2. Pirkani, A., Norouzian, F., Hoare, E., Cherniakov, M., & Gashinova, M. (2022). Automotive interference statistics and their effect on radar detector. IET Radar, Sonar & Navigation16(1), 9-21.
  3. F. Norouzian, A. Pirkani, E. Hoare, M. Cherniakov and M. Gashinova, "A Graphical Heatmap Tool to Analyse the Effects of Interference in Automotive Radar," 2020 IEEE Radar Conference (RadarConf20), 2020, pp. 1-6, doi: 10.1109/RadarConf2043947.2020.9266353.
  4. A. Pirkani, F. Norouzian, E. Hoare, M. Gashinova and M. Cherniakov, "Statistical Analysis of Automotive Radar Interference," 2020 IEEE Radar Conference (RadarConf20), 2020, pp. 1-6, doi: 10.1109/RadarConf2043947.2020.9266653.
  5. F. Norouzian, A. Pirkani, E. Hoare, M. Cherniakov and M. Gashinova, "Automotive Radar Waveform Parameters Randomisation for Interference Level Reduction," 2020 IEEE Radar Conference (RadarConf20), 2020, pp. 1-5, doi: 10.1109/RadarConf2043947.2020.9266375.
  6. A. Pirkani, F. Norouzian, E. Hoare, M. Cherniakov and M. Gashinova, "Analysis of Automotive Radar Interference in Spatial Domain," 2021 21st International Radar Symposium (IRS), 2021, pp. 1-8, doi: 10.23919/IRS51887.2021.9466171.
  7. F. Norouzian, A. A. Pirkani, E. G. Hoare, M. Cherniakov and M. Gashinova, "Characterization of the Effect of Low Pass Filter Response on the Interference in FMCW Automotive Radar," 2021 21st International Radar Symposium (IRS), 2021, pp. 1-6, doi: 10.23919/IRS51887.2021.9466189.
  8. A. Pirkani, F. Norouzian, E. Hoare, M. Cherniakov and M. Gashinova, " Automotive Interference Suppression in MIMO and Phased Array Radar," 2021 European Radar Conference (EuRAD), 2021.
  9. A. Pirkani, S. Cassidy, F. Norouzian, M. Gashinova and M. Cherniakov, "’Doppler Beam Sharpening for Enhanced MIMO Imagery in the Presence of Automotive Interference," 2021 European Radar Conference (EuRAD), 2021.

Research team