Spread Spectrum Implementation

Thread Starter

imzack

Joined Nov 3, 2010
73
I am struggling understanding Spread Spectrum, namely Direct Sequence Spread Spectrum (DSSS)

After reading ALOT of material, things are still not clear for me.

This article did provide some help:
https://www.allaboutcircuits.com/te...anding-spread-spectrum-modulation-rf-systems/

But most books just say that the "Data" is multiplied in some manor by a "Spreading Code".




Is the "Data in" Normally a On-Off Keying Signal?

Is the spreading code always multiply the "Data" by either '1' or '-1' as the article above suggests?

How does the rate of the spreading code, effect the spreaded bandwidth?


And what actual hardware devices are used to accomplish such a task?
A mixer isn't going to 'invert' the signal... right?
Do I have to use some type of dual microwave transistor pair to accomplish this?



I have a lot of questions, and every book seems to gleam over these questions I find most important.

Can anyone with extensive knowledge on this shed some light on this?


Thank you!
 

DickCappels

Joined Aug 21, 2008
10,152
Imzack wrote:

"Is the "Data in" Normally a On-Off Keying Signal?"

Yes, it is usually binary data. It can be done by modulating analog signals in a four quadrant multiplier or similar like when demodulating double sideband radio signals but with a despreading code rather than an oscillator.

Is the spreading code always multiply the "Data" by either '1' or '-1' as the article above suggests?
Yes, you can think of -1 and 1 as meaning invert the incoming signal or don't invert it.

How does the rate of the spreading code, effect the spreaded bandwidth?
See the article below for the formulas.


And what actual hardware devices are used to accomplish such a task?
See the article for an example.

A mixer isn't going to 'invert' the signal... right?
It inverts or doesn't depending upon the phase of the despreading code at the instant you compare the input with the output.

I am not usually the type of person to say "Go away and read this." but your questions would require a couple thousand words plus some pictures to answer. The article at the link below will answer all of your questions though like me, you might have to read it a couple of times until it all makes sense.

Everything you wanted to know about DSSS. An excellent article, maybe that is why it is still popular 25 years after its first publication.
https://www.tapr.org/ss_g1pvz.html

It would be helpful to review Barker codes, though other coding sequences can be used.
https://en.wikipedia.org/wiki/Barker_code

The nice thing about a good sequence is that if you multiply it with an asynchronous signal, like an interfering signal, the average value after the mixer (output of the low pass filter or integrator) is close to zero, but in the presence of a synchronous signal the output of the low pass filter is maximum.

An example of how simple the electronics can be is this impulse radio detector.
upload_2018-10-4_12-27-6.png
 

Thread Starter

imzack

Joined Nov 3, 2010
73
Thank you so much for the info DickCappels!

I have been reading, and re-reading.... As some things make more sense, more questions seem to arrise. (Although, I seem to be heading in the correct direction due to your provided sources)

Understanding Carson's rule better, is also helpful.... Is there a Carson's rule that applies to Phase Shift Keying?

Also, another thing that I cannot get my head wrapped around still, is that people say you can 'hide the signal UNDER the noise floor'

This makes fundamentally no sense to me, as I see the noise as a barrier you cannot 'dig' into. Like a cement floor, your signal must add to the overall power.


Maybe asking a question would clear things up....
Lets say I have a Direct Sequence Spread Spectrum Transmitter/Source. As I walk towards or away from the source, how would my spectrum appear? Would my noise floor increase or decrease as I walked towards and away from this source?

How can you say the signal is "Under" the noise floor? This make me want to think that if this was the case, you could have every user in the world transmit their own DSSS on a fixed bandwidth without issue, but according to Shannons Capacity theorem, and common sense, this does not appear to be the case.


Thanks
 

WBahn

Joined Mar 31, 2012
29,979
Thank you so much for the info DickCappels!

I have been reading, and re-reading.... As some things make more sense, more questions seem to arrise. (Although, I seem to be heading in the correct direction due to your provided sources)

Understanding Carson's rule better, is also helpful.... Is there a Carson's rule that applies to Phase Shift Keying?

Also, another thing that I cannot get my head wrapped around still, is that people say you can 'hide the signal UNDER the noise floor'

This makes fundamentally no sense to me, as I see the noise as a barrier you cannot 'dig' into. Like a cement floor, your signal must add to the overall power.


Maybe asking a question would clear things up....
Lets say I have a Direct Sequence Spread Spectrum Transmitter/Source. As I walk towards or away from the source, how would my spectrum appear? Would my noise floor increase or decrease as I walked towards and away from this source?

How can you say the signal is "Under" the noise floor? This make me want to think that if this was the case, you could have every user in the world transmit their own DSSS on a fixed bandwidth without issue, but according to Shannons Capacity theorem, and common sense, this does not appear to be the case.


Thanks
This is a common misconception -- that recovering signals below the noise floor should somehow be impossible.

It all has to do with being able to average the noise away and let the signal component build up over time.

As a real simply example, imagine generating millions of random values that are normally distributed about zero with a standard deviation of 1000. Now take your data and let each bit last for a hundred thousand samples and if the bit is a 1 you add 10 to the value and if it's a 0 you don't add anything. You now have a small signal embedded in a much large amount of random noise. But now average the data over a hundred thousand samples and you can start seeing the effect of the data and the noise gets suppressed into the background.
 

Thread Starter

imzack

Joined Nov 3, 2010
73
I'm sorry WBahn, but your example doesn't make sense to me.

All I can visualize, is adding power to the noise floor, so now you have a overall spectrum that has a overall higher noise power/ Higher noise floor after broadcasting your DSSS signal.

"Maybe asking a question would clear things up....
Lets say I have a Direct Sequence Spread Spectrum Transmitter/Source. As I walk towards or away from the source, how would my spectrum appear? Would my noise floor increase or decrease as I walked towards and away from this source?"
 

DickCappels

Joined Aug 21, 2008
10,152
It raises the noise floor as far as what an am receiver would see.

The source is transmitting noise, but not just any noise, this noise is a specific pattern, usually referred to as pseudonoise (It looks a lot like noise but it is a specific pattern). The overall input signal is demodulated by a copy of this pseudonoise signal. When the phase of the pseudonoise signal at the receiver matches up with the pseudonoise at the transmitter the signal used to modulate the pseudo noise at the transmitter is demodulated back to its original form.

Since all of the background noise is not synchronized with the pseudonoiseot remains noise, which tends to average to zero, so after the demodulator there is an integrator or low pass filter that causes the interfering noise to become much smaller.

Before demodulation with the pseudonoise, the signal looks like the blue trace in the image below.
upload_2018-10-11_13-56-21.png

After the pseudocode is used to demodulate the signal, it looks like the blue trace in the image below.
upload_2018-10-11_13-57-29.png
A little more filtering and slicing makes this into a real digital signal.

Incidentally, in practical applications the pseudonoise sequence is considerably longer than the 7 bits shown here.
 

Thread Starter

imzack

Joined Nov 3, 2010
73
Thank you again for your reply.


It raises the noise floor as far as what an am receiver would see.
So the transmitter is indeed transmitting 'noise' onto the noise floor, thus raising the overall noise floor.

Then why do people say that a signal is hidden "UNDER" the noise? Making it sound like no power is been added to the noise floor?





Let me see if I can summarize what you explained, and you can correct me where I am mistaken.

The pseudonoise((PN-Code Xor'd w/ Data) * Carrier) is broadcasted by the transmitter, thus raising the noise floor of the wide band transmission channel.
On the receiver end, it already knows the Digital PN Code (Spreading Sequence Code).
The receiver takes a CW Tone (centered at the same center freq.) and mixes this w/ the already known PN Code.
This creates a spread noise spectrum on the receiver, but just without the data (PN Code only * Carrier)

We then do some type of Auto-Correlation between the two psuedonoise sources (The Transmitted(w/ Data) & The Receivers internally generated psuedonoise signal(without data) )




Your screenshots make some sense, but the bottom screenshot doesn't seem to appear to be demodulating or recovering any beneficial data, other than what you already had (via your already supplied PN Sequence)


Sorry for the length of this post, but I think I am slowly getting some aspects, while still being stuck on others. I do however really appreciate your assistance in guiding me through this.


Thanks
 

DickCappels

Joined Aug 21, 2008
10,152
It is clear that you are "getting it".

In post #7 imzack wrote (responses in black):
The pseudonoise((PN-Code Xor'd w/ Data) * Carrier) is broadcasted by the transmitter, thus raising the noise floor of the wide band transmission channel.
On the receiver end, it already knows the Digital PN Code (Spreading Sequence Code).
The receiver takes a CW Tone (centered at the same center freq.) and mixes this w/ the already known PN Code.

This is a good place to note that there are many variation in the actual arrangement of the circuit blocks. Some demodulation from spread signals can be done in an RF stage in the receiver and sometimes it is done at baseband.

In some cases balanced mixers of one sort or another are used in places you mention XOR gates.

This creates a spread noise spectrum on the receiver, but just without the data (PN Code only * Carrier)

Not quite. In mixing of the CW signal with the incoming signal results in data modulated by the pseudonoise (hereafter PN), the carrier is supposed to be removed by the mixing of the and any interfering signals or noise being mixed down to baseband signals, subsequent low pass filtering will remove the any remaining RF.

This baseband signal is at this point composed of the spread data modulated by the PN. Running this through another balanced modulator (or mixer) with one input of the mixer being the modulated PN and the other being the local PN followed by a low pass filter gives you back your original PN.

This last demodulation step reverses or doesn't the signal in time with the PN, though chopping up any noise signals as a sequence of inverted and uninverted versions of the noise. The low pass filter on the output of this mixer is low passed so the inverted and uninverted bits of unwanted signal average toward zero.

For a better understanding of what happens to the noise, see the discussion of Process Gain in the James Vincent paper under the Spread spectrum terminology heading. It is about 1/3 of the way down this page
https://www.tapr.org/ss_g1pvz.html.

The bottom image in my previous post shows that the PN sequency can be recovered from the noisy signal. There are probably many ways in which to impress data on the PN The one most frequently seen (by me at least) is to invert the data or not as a function of whether the modulating PN is a 0 or a 1. Another is to transmit the PN signal or not, on-off coding.

As for the receiver finding the correct PN timing for the demodulator, a phase detector or PLL may be used. James Vincent described a novel Early-Late phase detector that works remarkably well.
 

Thread Starter

imzack

Joined Nov 3, 2010
73
DickChappels, you sir are a saint.

Thank you for having the patience to work with me; in helping expand my knowledge on this subject.

I by no means understand everything, but I believe you have armed me with alot of things/ideas/topics to start reading up on to help me continue learning on this subject.

I know now how the PN code is used to increase the spectral bandwidth of the original data. And typical spread spectrum systems typically use PSK, to make a uniform spectral 'noise' over a broad band.

The signal isn't hidden under the noise, but itself looks like noise, since it is so spread spectrum, and relatively low power (where it can be overlooked/hidden)

You have linked to good sources, that explain how you might go about implementing such modulators or demodulators, and even systems.

Again, I don't want to keep this thread going on indefinatley, but you have given me alot of good info, where I was hitting a brick wall. This info will help me move forward in continuing learning on this subject.

Thank you again!
 
Top