TIDEP-01021: Radar - MIMO & Beam-steering

Cesar,

• Thanks for your organized response.
• Further, I own three companies that buy from TI products worth over $150K annually. For that we expect responses to our questions every time we ask. We, not you, determine whether the issue is resolved or not. If you, rudely, think we should read before we ask, you should put this complaint to your supervisor to us, not to us. You should also ask him to teach you how to talk more courteously to your clients. In the mean time you can address our further Questions and keep your nontechnical advices to yourself;

1. The document “Design Guide: TIDEP-01021 Beamsteering for Corner Radar Reference Design” refers us to the document “Programming Chirp Parameters in TI Radar Devices”. The 1^st document states, in Page 6, that angular resolution will be improved by a factor of 3. The 2^nd document also clearly states (in Equations 7 & 8) that using MIMO system will result in virtual antennas which will improve resolution by a factor of 3. If we employ three Tx Antennas (instead of 1), does MIMO improve angular resolution by a factor or 2 or 3? Or does that factor depend on the mechanical configuration of the Tx antennas? If so which mechanical configuration results in a factor of 2 and which mechanical configuration results in factor of 3?

1. If our system has 3 Tx antennas and 4 Rx antennas and we employ beam-steering with a steering step of ~5 degrees à we should be able to distinguish two Objects one at an angle of 45 degrees and the other at an angle of 50 degrees (for example). This is because (as you said before in answering Q6 of our previous communication) at each of these angles we obtain the maximum possible resolution for the case of 1Tx antenna and 4 * Rx Antennas. That is, we eliminate the factor of 1/cos (40) in case of beam-steering. This is effectively angular resolution of ~5 degrees. At the same time if we employ MIMO and we end with 12 virtual Rx antennas then the resolution is 15 degrees, and at angle of 40 degrees we have to apply the factor of 1/cos(40) to obtain the effective resolution. In other words, it might harder to distinguish objects that are 5 degree apart in that MIMO case. In third words, we might be able to distinguish more objects (that are arranged in an angular direction) when we employ beam-forming and subsequently beam-steering. We accept that in between the steered angles the resolution might be really bad if compared with the case of MIMO and virtual antennas. Correct?

Regards

Sameh Refaat

Thank you for your feedback.

q1

The document “Design Guide: TIDEP-01021 Beamsteering for Corner Radar Reference Design” refers us to the document “Programming Chirp Parameters in TI Radar Devices”. The 1^st document states, in Page 6, that angular resolution will be improved by a factor of 3. The 2^nd document also clearly states (in Equations 7 & 8) that using MIMO system will result in virtual antennas which will improve resolution by a factor of 3. If we employ three Tx Antennas (instead of 1), does MIMO improve angular resolution by a factor or 2 or 3? Or does that factor depend on the mechanical configuration of the Tx antennas? If so which mechanical configuration results in a factor of 2 and which mechanical configuration results in factor of 3?

a1

Yes, the angular resolution depends on the mechanical configuration of the Tx antennae. The following appnote will help understand the MIMO radar.

In order to increase the angular resolution the TX antennae must be placed in the same line as shown in Figure 4

So when Tx are lined up, a system with 2TX/4Rx will result in virtual array of 8 Rx which will improve the angular resolution by a factor of 2. A system with 2TX/4Rx will result in virtual array of 8 Rx which will improve the angular resolution by a factor of 3.

The AWR1843BOOST antennae is similar to what is shown in Fig6. There are 2TX lined up in azimuth so the angular resolution is improved by a factor of 2 in azimuth

q2

If our system has 3 Tx antennas and 4 Rx antennas and we employ beam-steering with a steering step of ~5 degrees à we should be able to distinguish two Objects one at an angle of 45 degrees and the other at an angle of 50 degrees (for example). This is because (as you said before in answering Q6 of our previous communication) at each of these angles we obtain the maximum possible resolution for the case of 1Tx antenna and 4 * Rx Antennas. That is, we eliminate the factor of 1/cos (40) in case of beam-steering. This is effectively angular resolution of ~5 degrees. At the same time if we employ MIMO and we end with 12 virtual Rx antennas then the resolution is 15 degrees, and at angle of 40 degrees we have to apply the factor of 1/cos(40) to obtain the effective resolution. In other words, it might harder to distinguish objects that are 5 degree apart in that MIMO case. In third words, we might be able to distinguish more objects (that are arranged in an angular direction) when we employ beam-forming and subsequently beam-steering. We accept that in between the steered angles the resolution might be really bad if compared with the case of MIMO and virtual antennas. Correct?

a2

No, this is not correct. The steering step does not define the angular resolution. You are correct the factor 1/cos(40) is eliminated. However the angular resolution is still 2/N where N is the number of number of Rx equispaced antennae (as shown in the MIMO Radar AppNote).

So for the AWR1843BOOST antennae the angular resolution will be 30deg when using beam steering

Thank you

Cesar

Cesar,

A1 for Q1 is clear.   Thank you.

Q2. If we forget about the strict definition of the term "angular resolution" for now....would you agree that beam-steering (if we set the angle step to 5 degrees) will be a better way to distinguish 2 objects that are 5 degrees apart (for example an object that is located at an angle of 40 degrees and the other that is located at an angle of 45 degrees) as compared to any attempt to distinguish between these two objects using MIMO solution that employ 2 Tx Antennas and 4 Rx antennas? That MIMO solution and the resulting 8 virtual antennas will have difficulty distinguishing objects that are 5 degrees apart. Correct?

Regards

Sameh

Q2.

If we forget about the strict definition of the term "angular resolution" for now....would you agree that beam-steering (if we set the angle step to 5 degrees) will be a better way to distinguish 2 objects that are 5 degrees apart (for example an object that is located at an angle of 40 degrees and the other that is located at an angle of 45 degrees) as compared to any attempt to distinguish between these two objects using MIMO solution that employ 2 Tx Antennas and 4 Rx antennas? That MIMO solution and the resulting 8 virtual antennas will have difficulty distinguishing objects that are 5 degrees apart. Correct?

A2:

We need to emphasize at this point that we are considering static objects. So, the objects can't be distinguished using their velocities.

The attempt to distinguish between the two objects that are 5 degrees apart is limited by the angular resolution of the system. The system can't distinguish between the two objects because there is not sufficient angular resolution. Both beam-steering and MIMO will not have sufficient angular resolution to distinguish between two objects 5 degrees apart. Both objects will be detected as a single object.

The advantage of beam-steering over MIMO is that if we have one object at 40 degrees for example. The SNR will be higher for beam-steering than MIMO.

Thank you

cesar

Casar,

Even If there the objects are moving in that specific case (I mean the case of two objects that are located at 40 degrees and 45 degrees)...where the objects are moving in a radial direction or in an angular direction with limited displacement around these  two location ....beam-steering will serve the purpose better than MIMO.....correct? 

Velocity will help us getting more information using Doppler effect and will allow us to improve the detection of both beam-steering case and MIMO case. The improvement will be equal in both cases. Correct?

Regards

Sameh

Hi,

If the objects are moving so that they can be separated in the doppler domain, than you are correct, the SNR will be better when using beam-steering.

If the objects are static than they will be detected as a single object with both beam-steering and MIMO. However the SNR will be better when using beam-steering.

Velocity will be very important in this use case since it may enable to detect the two objects. Without velocity information it will not be able to detect the two objects since the system does not have sufficient angular resolution

Does this answer your questions?

thank you
cesar

 Caser,

Thanks for the answer. It answers the question partially.

I will re-phrase my previous question;

You said earlier “We need to emphasize at this point that we are considering static objects. So, the objects can't be distinguished using their velocities.”

Here I am trying to establish that beam-steering might remain superior option in some cases where the object(s) are moving and where these objects can be distinguished using their velocities.

If the object(s) are moving in a radial direction (ie towards or away from the radar) then the extra information obtained based on velocity will be the same in case of both cases of MIMO and beam-steering and hence the ability of both techniques to distinguish objects will not be influenced by the radial motion of the object(s). Correct?

On the other hand, if the object(s) are moving in an angular direction (ie perpendicular to the line joining the object(s) and the radar) --> beam-steering (that has 5 degrees step) might and might not be better option than MIMO (with 8 virtual antennas) in terms of their ability to distinguish/separate objects. Which technique is better that will depend on the object(s) tangential/angular velocity. Correct?

Naturally real motion is a combination of radial and angular motion.

Regards

Sameh 

Hi,

Yes, you are correct, if the objects can't be distinguished in the doppler domain, beam-steering is a superior option because it will provide more SNR.

If the objects are moving in radial directions such that they fall in distinct doppler bins, than theoretically they can be separated. However in order to be separated in doppler domain they must first have sufficient SNR to be above the noise level when the Range FFT is performed. This is why beam-steering is a better option. with MIMO there may not be sufficient SNR

If the objects are moving in tangential directions than it is not possible to separate them based on velocity information. So if the goal is to separate the two objects, than beam steering is not an better option than MIMO. Both will fail. However in the Range FFT the peak will still have a stronger SNR for beam steering than for MIMO,

Thank you

Cesar