How will I model the signal and the array manifolds in case of near field scenario? Also, here only azimuth angle is shown, how will the near field signal model look like incase of both azimuth and elevation angle
$\begingroup$
$\endgroup$
2
-
$\begingroup$ For a ULA, the far-field approximation is essentially that the range from the $n$-th element to the source is $R_{n} \approx R_{0} + \frac{nd\sin(\theta)}{\lambda}$ where $R_{0}$ is the range from the reference element to the source. So, we can model the difference in range as a uniform phase shift over a narrow band. This assumption is violated in the near-field, so your phase shift would be non-uniform across the elements. $\endgroup$– BaddioesCommented Nov 8 at 14:49
-
$\begingroup$ Real quick...why do you want to do stuff in the near-field? $\endgroup$– EnvidiaCommented Nov 8 at 14:59
Add a comment
|
2 Answers
$\begingroup$
$\endgroup$
4
That depends on the array; generally, in near-field, your receiver interacts with the transmitting elements, so you simply "break" the array geometry as something consistent of individual independent antennas with fixed phase centers. So, the very basic assumptions of radar is not fulfilled anymore! You will very likely have to resort to electromagnetic simulation (typically, a finite-elements methods & method of moments).
Also, hm. We're talking about radar here. If you have something that you call an "array", then that typically is in the order of a wavelength in size. If you are in near field, you're nearer than 2 wavelengths. Is this really a case you care about?
- By necessity, to measure nearby objects, you want a high bandwidth for your radar to get good resolution. (If your resolution is, say, 1m, then why would you care how things change if things are in near field, if near field is much smaller than 1m?)
- To get high bandwidth, you need a high carrier frequency (because carrier frequency needs to be much larger than bandwidth, or you won't be able to build an antenna).
- If you have a high carrier frequency, then your wavelength is very small (it is thus much smaller than your resolution due to bandwidth).
- If your wavelength is very small, then your near-field (which typically ends latest 2 wavelengths from the antenna) is also only very close by
- if your radar target is so close by, you have (see 1. point) misdesigned your system, because you can't really resolve that distance anyways.
As a practical example: Maritime radars (so, things that detect radar targets kilometers away!) might be operating in the X band (so, 8 to 12 GHz). The near field is hence somewhere less than 4cm away from the antenna. If a ship is less than 4cm from your ship's radar antenna, you should put on a life vest. Whether it's exactly 4cm or less doesn't matter. Prepare to swim.
Another example: Automotive radars are in the 24 GHz, 60 GHz or 77 GHz band. That means far field already starts 25 mm, 10 mm or 7.8 mm away from your radar antenna array. If a car is less than 25 mm from your radar sensor, and your car is moving at any significant speed, it should have started emergency breaking a long time ago, and the air bag should be starting to inflate by now, and hopefully the car has already called for an ambulance (by law, new cars in Europe have to be able to call for help in a crash, autonomously).
If so, you need near-field detection to go for a different kind of technology than radar, I'm afraid. I'd recommend you look into RFID detectors, passive sensor interrogation techniques, and material science done with microwave interacting with the material, detuning the system.
answered Nov 8 at 13:54
-
$\begingroup$ For a fixed size antenna, as the wavelength decreases the far-field distance increases. $\endgroup$– EnvidiaCommented Nov 8 at 15:58
-
$\begingroup$ yes, but you don't go with the same antenna size (and array spacing) as you increase the frequency. that way would lie grating lobes! $\endgroup$ Commented Nov 8 at 16:06
-
$\begingroup$ True, but considering just the far-field distance, it is important to note that it's actually the ratio of the physical size relative to the wavelength that defines it, grading lobes or not and applies to any antenna in general. $\endgroup$– EnvidiaCommented Nov 8 at 16:19
-
$\begingroup$ @MarcusMüller, thank you for the explanations. I was curious because I saw a paper on 'A Comprehensive Review of Direction-of-Arrival Estimation and Localization Approaches in Mixed-Field Sources Scenario'. But yes from practical view point, it's not useful for radars. $\endgroup$– ananyaCommented Nov 9 at 18:13
$\begingroup$
$\endgroup$
0
Much of the theory and processing of radar signals assumes that the fields meet the far-field assumption. The far-field is the minimum distance where the range-dependent component of the fields is negligible, and the fields can be seen as a "plane" moving through space, hence the term "plane-wave".
The minimum far-field distance is not dependent on the wavelength $\lambda$ alone, but also the largest dimension $D$ of the antenna. In the case of a square array for example, the physical diagonal would be the distance used for $D$. So the minimum far-field range is actually defined by a ratio of $D$ and $\lambda$. There are other criteria that could be used but the commonly used approximation is
$$R_{min} = \frac{2D^2}{\lambda}$$
So for a fixed physical antenna size, decreasing the wavelength increases the minimum distance required to be in the far-field. When using an array, you need to consider their spacing to meet the "spatial sampling" criteria to avoid grating lobes, but the far-field expression is a good approximation for any antenna if that's all you care about.
So trying to do traditional signal processing techniques on array signals is a non-starter if you're considering the near-field. Looking at how steering vectors are defined, the are a function of angle only because of the far-field assumption. Trying to perform certain processing in the near-field would require intense characterization of the near-fields of your particular antenna, which is dependent on range, which is competing against the reality that radars are trying to estimate the range of a signal with beamforming and the other downstream processing!
The near-field is not useless however. Near-field antenna ranges are used to capture near-field data, and through a mathematical transformation, can yield the far-field pattern. This is very useful for antennas operating at frequencies that would make the far field very large (tens to hundreds of meters).
