Digital signal processing

And they do not run on the game thread.

They run on their own dedicated, heavily optimized digital signal processing thread, a DSP thread, that operates natively at the audio sample rate.

Okay, so a flow graph requires a complete rewiring of how a developer thinks about logic.

The sources compare it to a modular analog synthesizer.

It is a highly optimized, reusable, self-contained mathematical function.

You build the logic once, save it as a patch, and then you can drag and drop that patch into any master MetaSound source as a single clean node.

Which dramatically reduces visual clutter and prevents you from having to rewrite the same DSP logic a hundred times over.

It saves hours of recoding.

Find fall your way at Ross.

Last time, um, we, uh, started talking about the most general type of, um, FMCW radar, which is the MIMO radar, right? Uh, that's the case where you have multiple transmitter antennas, multiple receiver antennas, and then it enables you to, uh, this, this antenna diversity basically enables you to determine the, the direction of arrival of targets with some, uh, clever DSP.

Uh, so we started by introducing this idea of the virtual antenna array, and this is a purely mathematical construction.

Basically, uh, if you have a MIMO array which has, uh, N sub TX transmit antennas and N sub RX receive antennas, so physically you only have NTX plus NRX antenna elements, right? Um, but mathematically we showed that it is equivalent, and by mathematically equivalent, I mean, uh, essentially the signal that you receive from it, uh, is equivalent to a, a SIMO array, single input multi-output, where you have just one transmitter and you have P receivers, where P is equal to the number of transmitters times the number of receivers in the original antenna array.

So the tilde is for the, uh, virtual array.

Yeah, you can, you can cross-check.

Uh, but the point is, uh, now we can, uh, basically do all of our DSP, uh, in, in the, in the virtual, uh, um, in the s- in the signal model that we define on the virtual array.

Uh, so the way it looks like is, like, let's say physically you have a two-by-three MIMO array.

Then you apply some filtering to get rid of out of band noise and just keep the frequency content where you know there is signal.

And then you digitize it through an ADC, and then you do some processing in DSP to estimate your range and bearing angle, and we'll talk about the processing.

Now note that this is kind of like a generic block diagram.

Specific systems might be different from this in certain aspects.

There's the oscillator.

There's the waveform generator.

and then the amplifiers on the transmit and the receive side, filter, ADC, and possibly some DSP on board to basically process the signals.

Some of them you don't do the DSP on board and just give you, you know, digitized signals and then you can have a separate microcontroller or some compute elsewhere that does the processing for you.

It can even be fully implemented in analog.

So you might not even, uh, need to use like a ADC and a DSP.

Uh, um, you can, you can do it much simpler.

But it comes with all, uh, uh, sorts of, uh, uh, you know, um, um, um, uh, caveats in terms of, uh, specifically the error in ranging has, has, has dependency on the amplitude of the received echo and also on the angle of where the target is.

Where is this coming from? Well, we know rho is, uh, basically the time of flight, um, times the speed of light over two, and then we have just plugged in, uh, uh, uh, for, for tau, we have, we have plugged in, um, this expression, which is delta F over B.

So that's our time of flight, okay? So really, the magic or the power of, of, of, of this idea of CTFM or FMCW is that we have converted a, um, range estimation problem to a frequency estimation problem.

So now we just need to, uh, estimate the frequency of the signal, uh, in, in DSP.

And, uh, as I said last time, there is tens or possibly hundreds of algorithms, uh, for, for frequency estimation.

It doesn't use a cheap radio tuner.

It utilizes custom DSP digital signal processing sweep architecture.

Right, and DSP changes the game entirely.

How so? Break down DSP for us.

It means the device is using microprocessors to analyze and manipulate the audio signal mathematically in real time.

Instead of just blindly scanning analog dials.