[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Discuss-gnuradio] Confusion about Sample rate

From: Kyeong Su Shin
Subject: Re: [Discuss-gnuradio] Confusion about Sample rate
Date: Sat, 1 Jun 2019 17:40:55 +0000

Hello Jiji,

Sampling rate of the system can affect the performance of a such system (depending on the performance matrices that you use). Here's a simple example:

Let's say someone is transmitting WiFi signals over the air. Due to the multipath fading, some portion of the spectrum is more attenuated than the other part of the spectrum (in the frequency domain). If your receiver is sampling at a slow sampling rate (significantly lower than the Nyquist rate), and happens to be at the attenuated side of the spectrum, then it may miss the transmitted signal. That would not have happened if the receiver sampled at the full rate.

Now, the reality is not that simple (for one thing, the effective bits of USRPs may increase when you configure it to sample at a lower rate, because it can actually sample the signal at the full rate and then decimate on the FPGA; that is how most real-world receivers work), but various factors like this can effect the probability of detection, false alarm, or the detection speed of the spectrum sensing system. 


Kyeong Su Shin

보낸 사람: Jiji Varghese <address@hidden> 대신 Discuss-gnuradio <discuss-gnuradio-bounces+address@hidden>
보낸 날짜: 2019년 6월 2일 일요일 오전 12:35:40
받는 사람: Marcus D. Leech; address@hidden; address@hidden
제목: Re: [Discuss-gnuradio] Confusion about Sample rate

Thank you for the responses.

I was planning to use USRP B200 which allows a BW of 56MHz.
Yes. I just want to sense the WIFI spectrum. I came across various sensing techniques like energy sensing, cyclostationarity, covariance based  etc. Hope I am right. But I didn't get how performance tradeoffs affect sample rate?

J. Varghese

On 12:09PM, Sat 01-Jun-2019 Marcus D. Leech <patchvonbraun@gmail.com wrote:


> >

> On 06/01/2019 02:03 AM, Kyeong Su Shin wrote:


> >>

>> To whom it may concern:
>> While I mostly concur with Marcus's response, I would like to note that maybe you can get away with a lower sampling rate (lower than the Nyquist rate) since you are working on spectrum sensing (not on decoding the full WiFi signal). 
>> Actual sampling rate needed by your application will depend on the spectrum sensing algorithm and various performance factors that you want to use.
>> Regards,
>> Kyeong Su Shin
> Agreed.  I had missed the "for spectrum sensing" part.  There are a number of algorithms possible, with various performance trade-offs
>   possible.


> >> ________________________________

>> 보낸 사람: Marcus D. Leech <patchvonbraun@gmail.com> 대신 Discuss-gnuradio <discuss-gnuradio-bounces+ksshin=postech.ac.kr@gnu.org>
>> 보낸 날짜: 2019년 6월 1일 토요일 오후 12:28:49
>> 받는 사람: discuss-gnuradio@gnu.org
>> 제목: Re: [Discuss-gnuradio] Confusion about Sample rate
>> On 05/31/2019 11:21 PM, Jiji Varghese wrote:


> >>>

>>> I am trying to store WiFi signals samples in a file using file sink block, using GNU radio, for spectrum sensing. I know that a single WiFi channel has a bandwidth of 20-22 MHz. So should I use the sample rate in GNU radio block to 40e6 - 44e6 or can I use a lower value? Can I use 1e6 as sampling rate?
>>> What is the lower bound for the sample rate in GNU block? 
>>> I am confused about giving the sample rate.
>>> J. Varghese
>> A couple of different issues here:
>> GnuRadio *generally* deals with *complex samples*.  Complex  sampling satisfies the Nyquist Sampling Criterion in a way that allows you to
>>   run those samples at a rate == bandwidth (more or less).
>> Gnu Radio doesn't care, per se, about sample rates, and will pass your request on to the underlying *HARDWARE*.  The hardware will, most
>>   definitely have limits and policies with respect to sample rate, and to answer the question "what sample rates are allowed" you need to refer
>>   to the hardware you're using.
>> If you don't understand what the Nyquist Sampling Criterion is, you should definitely spend some time researching that, and how it relates
>>   to the kinds of things you might be wanting to do with samples, and why 1e6 is not an appropriate sampling rate for a signal with an
>>   information bandwidth of 22MHz.
>> _______________________________________________
>> Discuss-gnuradio mailing list
>> Discuss-gnuradio@gnu.org
>> https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
> _______________________________________________
> Discuss-gnuradio mailing list
> Discuss-gnuradio@gnu.org
> https://lists.gnu.org/mailman/listinfo/discuss-gnuradio

reply via email to

[Prev in Thread] Current Thread [Next in Thread]