Wideband digital beamforming at subarray level

[berraco]

digital subarray model

Hi, all.
This post can probably be placed at antenna forum, but I think it is a DSP matter more than a RF/MW.
I'm interested in digital beamforming for radar phased arrays. When using narrowband signals, digital beamforming can be mainly achieved by FFT processing. But, what about wideband signals? And what happens if the array is very large and the digitization is performed at subarray level?
What kind of processing must/can be done? Beamspace processing?

Have any of you references (papers, books) about these situations (I mean, wideband radar with DBF at subarray level)?

Thanks in advance.

 

[eblue]

Hi berraco,

Look for ESPRIT and MUSIC super resolution algorithms. They are usualy used for that kind of processing.

 

[Beetroots]

Oversample the signal, use CORDIC processing for fine phase control, and perform fast RX/TX signal correlation (via FFT's).

 

[berraco]

Hi berraco,

Look for ESPRIT and MUSIC super resolution algorithms. They are usualy used for that kind of processing.

I'm not really sure, but I think those algorithms are aimed to DOA estimation. I have done a (not extensive) search and always found articles/references in that sense. So they could be good for tracking and for DOA interference estimator but what about a fixed beam (for example in a SAR system)? Can they be used? Can achieve good SLL (30 or 40 dB) and narrow beamwidth?

 

[satellite]

Dear berraco,

if you are involved in wideband digital beamforming, you will find useful to have a look to what have been done by decades on SONAR processing , where for its nature, the involved bandwidth covers several octaves.

The starting point to understand the principles of wideband beamforming is the sum and delay architecture.
When the signal band is small the time delay can be replaced by a phase delay, and this is the principle of many RF DBF architectures.
When this assumption is no more valid, true time delay must be implemented in a digital fashion (e.g interpolation).

Let me know if you need any specific clarification.

Bye, Satellite

 

[berraco]

Hi, satellite.

I've been checking SONAR systems and the approach about true time delay,
interpolation, is what I was thinking to use. First a zero padding to achieve a little more time resolution and then an interpolation filter, probably a polyphase filter will be the best choice, am I right?

But I have some more doubts. After interpolation, the next step is to handle with the digitized and time steered outputs of each channel, once per subarray. They are (originally) beams, so I expect to use beamspace processing. But actually they only are complex numbers, so I don't know. If I had an adaptive system, I think a MVDR algorithm will be a good choice. If not adaptive, e.g. SAR system fixed beam, what should I use? Must I create a set of beams and then combine them linearly? If this is the correct approach, how to define the problem and how to solve it? I'm really lost in this part.

 

[satellite]

Dear berraco,

the best beamforming imlementation will depend on the application you have in mind.

For multiple beam formation and adaptive systems, beamspace approache best exploit the nature of the signal with contemporay creation of all the required beams (in multibeam) or exploitation of the focused nature of desired and jamming signals (adaptivity).

Beamspace approach is also well suited for sub-arraying.

Nevertheless, if you are considering subarraying for reducing the digital section complexity, you have to split the problem in what concern the application and the complexity reduction.

In the simplest delay and sum architecture, subarraying shoud be thought as a hyerarchical division of the sum problem in an analog pre-summing and in a digital post-summing.

Sorry but my baby is crying, I will be back later! Satellite

 

[vanvu]

Dear berraco,

if you are involved in wideband digital beamforming, you will find useful to have a look to what have been done by decades on SONAR processing , where for its nature, the involved bandwidth covers several octaves.

The starting point to understand the principles of wideband beamforming is the sum and delay architecture.
When the signal band is small the time delay can be replaced by a phase delay, and this is the principle of many RF DBF architectures.
When this assumption is no more valid, true time delay must be implemented in a digital fashion (e.g interpolation).

Let me know if you need any specific clarification.

Bye, Satellite

Dear satellite,

You said that: When this assumption is no more valid, true time delay must be implemented in a digital fashion (e.g interpolation).

Pls tell me more about it ?

I'm looking for how to estimate the DOA with wideband signal.

Thanksss