AWR1443: FMCW radar angle resolution (MUSIC algorithm)

Tony Zhu

Part Number: AWR1443

I have read several TI training documents, and find that the angle resolution is determined by the number of receiving antenna.

The formula is : angle resolution = 2 / N, and the doa algorithm is based on angle-FFT.

However, there's some subspace algorithms, such as , MUSIC ,ESPRIT, and they were called super-resolution algorithms.

Does this mean these super-resolution algorithms can increase the angle resolution.

For example, if there is only 1 TX and 4 RX , in the angle-FFT, the resolution is 28.5deg.

While I choose MUSIC algorithm , will the angle resolution increase? What is the formula?

over 6 years ago

0 Anil Mani over 6 years ago

TI__Expert 4720 points

Hi

The performance of MUSIC/ESPIRIT (when the number of targets is correctly estimated,) is bounded by the 'Cramer Rao Bound' (for high SNR scenarios).

There is no closed-form expression for the 'Cramer Rao Bound' for multiple targets, however, it is fairly easy to calculate. You can use 'Fundamentals of Statistical Signal Processing: Estimation Theory (Kay, S. M.)' which has a section on how to calculate the bound for direction of arrival problems.

In our experiments, MUSIC/ESPRIT has fairly good performance when the number of 'virtual antennas' is high (say 40). For small number of antennas (say 4) we haven't seen good performance.

Regards
Anil

0 Tony Zhu over 6 years ago in reply to Anil Mani

Prodigy 20 points

Thanks, Anil

I can not figure out the relationship between MUSIC angle resolution(two targets with same range and speed) and CRLB.

Can you give me a brief introduction?

The MUSIC method :

Cramer Rao Bound:

Sensors

Sensors forum

AWR1443: FMCW radar angle resolution (MUSIC algorithm)