Communication & Signal Processing - Science topic

Hugh Lachlan Kennedy My implementation is attached and it's equivalent to use 'y = filter(B,1,x)' in MATLAB, where B is your FIR coefficients. I've tested it with some actual data and all results are the same.

Dear Fernanda Catelani ,

I know that my answer to your question is very late. As your question is a basic question I would like to give my opinion.

- The peak value of a sinusoidal wave can be found in time domain by discretizing the time domain signal and build a time sequence of amplitudes. By comparing the values we can locate the highest amplitude which can be considered as corresponding to the peak value. The value located will be most accurate when one one samples at the time of the peaks and also when increases the sampling frequency.

So there are two requirements to increase the accuracy:

- Aligning the sampling with peaks

- Increasing the sampling frequency

Let is assume that we transformed this signal to frequency domain.

It is so that the DFT is just the sampling of the frequency spectrum of the signal.

So one can locate a signal in the discrete frequency domain with the resolution of

fs/2N. One always hits a signal at the intended frequency +- fs/2N.

As for the Fourier transform of single sine wave with a time window N/fs, one gets sinx/x Fourier transform. with the sine wave at the center of this curve.

The discrete Fourier transform will locate the point of maximum amplitude with the resolution of fs/2N as i hinted previously.

So, there will be an error in locating the peak value either in the time domain and in frequency domain because of the discritization.

Best wishes

No, There are many different functions may yield the same FT image.

I will not repeat the answer of my colleague Navid. I would like to add some conceptual comments which may be useful for the issue.

Which channel model is dominant depends on the transmission media. If the media is homogeneous the line of sight model is dominant. If the medium contains large obstacles such as building then lognormal channel model will prevail. If the medium is scattering rich it will be multipath dominant with Rayleigh fading is the most suitable model. In case of making a study one can study the effect of such propagation models on the system performance.

As far the variation of the channel model with time represented by resource element, depends on the time variation of the channel. Generally the channel can be considered consatnt during the correlation time which is dictated by the Doppler shift.

Best wishes

It depends on the journal in which you will publish the paper

Best Regards Mohammed Saquib Khan

first of all, the factor of saftey is very important, if a car stops on the road it just sits there while if it flies it would just fall on people or houses, there is two technological “limitations“ that needs to be considered, one of them being what will you use to power the car ? if it was gasoline then you are talking about a considerably large jet engine, if it was electric then you would have two problems first being that the propellers would be large in size and would not generate much thrust, the second which is a technological limitation of storing electric power most commonly used method is li-po and li-ion batteries which is not verry efficient and have safty issues of self igniting if subjected to shock, overcharging, and short circuiting.

journal suggester of Springer and Elsevier will help. Please stay away from individuals asking for your paper to be emailed to them! Please avoid journals like IJSSE/IJERT and the likes. Verify the I.F claims of any journal by checking in the clavariate site and JCR.

dear,

try this program:

%"Author: Krishna"

clear N = 10^6 % number of bits or symbols rand('state',100); % initializing the rand() function randn('state',200); % initializing the randn() function % Transmitter ip = rand(1,N)>0.5; % generating 0,1 with equal probability s = 2*ip-1; % BPSK modulation 0 -> -1; 1 -> 1 n = 1/sqrt(2)*[randn(1,N) + j*randn(1,N)]; % white gaussian noise, 0dB variance Eb_N0_dB = [-3:10]; % multiple Eb/N0 values for ii = 1:length(Eb_N0_dB) % Noise addition y = s + 10^(-Eb_N0_dB(ii)/20)*n; % additive white gaussian noise % receiver - hard decision decoding ipHat = real(y)>0; % counting the errors nErr(ii) = size(find([ip- ipHat]),2); end simBer = nErr/N; % simulated ber theoryBer = 0.5*erfc(sqrt(10.^(Eb_N0_dB/10))); % theoretical ber % plot close all figure semilogy(Eb_N0_dB,theoryBer,'b.-'); hold on semilogy(Eb_N0_dB,simBer,'mx-'); axis([-3 10 10^-5 0.5]) grid on legend('theory', 'simulation'); xlabel('Eb/No, dB'); ylabel('Bit Error Rate'); title('Bit error probability curve for BPSK modulation');

Interested

Dear zakaria Masoud

Sorry for the late answer.

No I did not train the FFANN. I find the answer of the question. I did just a small mistake. With the specific order, when I generate each element (for example the m^th), I should not include that into next generated data (m+1^th ). The determined order makes the correlation in the series.

Far difference between SIW resonator and dielectric resonator. The loss of a resonator is dominated by copper loss which is not present in DR. Replace all metals of SIW by PEC then Qu will exceed even 5000.

Hello,

overall, your code is not bad, it just needs to be improved. I mean: 

- you didn't create a signal, but symbols. In order to simulate a single carrier signal, you just have to oversample (zero pading) and filter (root raised cosine for instance) the symbols.

- I don't understand why you multiply your channel by sqrt(P/2). If it is for a normalization matter, then it is wrong. You may remove it. However, your 3 taps channel is OK, the Rayleigh-ness of the channel is intrinsic to the function randn. In further  works, just take into account the delays and the gains of the taps. But for a first test your code is fine.  

- Then you can use the conv function to filter the signal by the channel.

- The SNR is usually defined as the power of the received signal over the noise variance. So for a given SNR : assess the power of the received signal P = sum(abs(x).^2)/length(x), then create a noise vector of power 1 : N1 = sqrt(0.5)*(randn(1,L)+1i*randn(1,L)), and finally create the noise vector with expected power: 

N = sqrt(P_N)*N1 with P_N = P*10^(-SNR/10)

- Finally, the signal is the sum of the transmitted signal over the channel plus noise N.

Regards

Thank you dear Luay Shihab . I think this is very clear resume on the compression. Now I understand better the compression process.

So on  practice or  implementation. My idea is to implement arithmetic compression which is an entropy compression (lossless). My test vector is a vector of 500 elements for example. Each element can take the value from 1 to 256. In my vector there will necessarily redundancies. And I can do compression with an arithmetic algorithm: the one proposed by: " Arithmetic coding for data compression I. Witten, R. Neal am\nd J. Cleary Communication of the ACM, 1987 Volume 30 Number 6 "

It works well but I would like to know is that there would not be another method more effective? Which would make it possible to have one with the same vector a better compression.

Ps: if you are interested I can share the part of the code that I modified and implement.

thank you.

Hi,

Probably the equalizer should reset each SNR loop. I did it through a workaround but you can check the MLSE structure and do reset each snr loop.

check this code, it works!

clc;clear;

close all;

ii=1;

snr_vec=-10:1:30;

for snr_i=1:length(snr_vec)

 

snr=snr_vec(snr_i);

 

N = 100000;

%chCoeffs = [0.986+0.5i;0.845;0.337;0.32345+0.31i;0.2];

chCoeffs = [0.5;0.00000000001];

mlse = comm.MLSEEqualizer('TracebackDepth',10,'Channel',chCoeffs,'Constellation',(1/sqrt(2))*qammod(0:3,4));

errorate = comm.ErrorRate;

 

 

Tx_Data = randi([1 2],N,1)-1;

modData = (1/sqrt(2))*qammod(Tx_Data,4,'InputType','bit');

chanOutput = conv(chCoeffs,modData);%filter(chCoeffs,1,modData);

chanOutput = chanOutput(1:end-length(chCoeffs)+1);

noise = (randn(1,length(chanOutput))+1i*randn(1,length(chanOutput)))/sqrt(2)/sqrt(db2pow(snr));

Rx_data = chanOutput+noise.';%awgn(chanOutput,snr);

equalized_Data = mlse(Rx_data);

% clear mlse

% mlse = comm.MLSEEqualizer('TracebackDepth',10,'Channel',chCoeffs,'Constellation',(1/sqrt(2))*qammod(0:3,4));

% reset(mlse)

demodData = qamdemod(equalized_Data,4,'OutputType','bit');

errorStats = errorate(Tx_Data,demodData);

% errorStats(2)

ber(snr_i) = errorStats(1);

save ber ber snr_vec;

NumofErr(snr_i) = errorStats(2);

% ii=ii+1;

clear;

load ber;

end

semilogy(1:length(snr_vec),ber)

grid on;

 

Dear Rupaban,

I suggest to you attach publiction and files in topics.

Best regards

What do you mean exactly?!

I saw the files. You should enter the equations in Matlab with commands such as sum,...

Thanks for your answers. I finally found something that helped me to visualize the general idea. In Blind Drift Calibration of Sensor Networks using Signal Space Projection and Kalman Filter they explain what is a 'signal subspace' and also how to construct the orthogonal projection matrix P. 

Thanks a lot

Dear Moahmed,

in OFDM we use one-tap equalizer, assuming the channel can be approximated by a single complex factor over the frequency range that a subcarrier occupies (layout of your subcarrier-bandwidths such that the channel can be assumed frequency-nonselective). This equalization is done by the complex multiplication.

Trellis Coded Modulation (TCM) which combines the coding and modulation together.

There are a lot of literature on TCM in 1990s.

Thanks Abdullah, tried to download it but it didnt work. Anyways, I am reading one relevant paper and hope i ll soon find the answer to my question 

Dear Mitun,

I suggest to you link and attached files in topics.

-Patent US8286048 - Dynamically scaled LLR for an LDPC decoder ...

Best regards

computing and comparing the RMS error

Variance is computed over many experiments for a random variable. An average value computed over a long period should be same as defined in your code. Do an experiment by  capturing complex Gaussian RV:  h=(1/sqrt{2}) * (N(0, 1) + j N(0,1))  for t=1, t=10, t=100, t=1000, and t=10000 times. Compute mean(h) and var(h) for t=1, t=10, t=100, t=1000, and t=10000. You will see that mean and variance converge to its specified value. In essence, instantaneous values can be greater/equal/less than the specified value but the average value will converge to the desired value.  Note that Rayleigh fading coefficient is abs(h).

It could be based on OFDM, or some other multicarrier scheme. OFDMA usually refers to a system with only one user per subcarrier, while all users in massive MIMO use all subcarrier. The separation is in the spatial domain instead. 

OFDM is a multiplex scheme, if the transmitted data series is a chaotic one, than the corresponding OFDM signal may have some chaotic properties. Without any restrictions on the modulated data, we cannot say that an OFDM signal is a chaotic signal. For example, if the data is a periodic one, then the OFDM signal is also periodic, It is obviously not chaotic.

Intuitively, you proposal is right. You need to make the question more clearly.

I work on the studies of stochastic resonance which is able to utilize noise to detect weak signals and moreover is suitable to short data sets.

 In LTE network, a UE measures two parameters on reference signal. Reference signal is transmitted with the OFDM symbol. OFDM symbol 0 and 4 in the slot are for reference signals. The following URL gives very good understanding of what you might be looking for. 

Dear Khalid & Lavish,

Many thanks for your reply. I already have many papers on this subject and the following dissertation is excellent. 

However, I face a problem in the implementation of the ICI estimation method especially the Linear Interpolation LI-CPE method which was mentioned in the in the paper that Lavish mentioned.

If you are speaking about the sampling characteristics when sampling ECG then the first question to ask is 'what problem are you solving?'.  Specifically, are you only interested in calculating HR or do you intend on detecting arrhythmias?  If arrhythmias: life-threatening, atrial, ventricular?  Do you want to work with people with pacemakers (single/dual)?  The answers then will drive your sampling rates of anywhere from 100 Hz to 500 Hz,  It also will drive what type of bandwidth filter you want to use.  Real-life ECG does not look like it does on TV hospitals shows:  it is full of noise.

The reference frequency has to be the same as the frequency of the signal to be measured. (Although some lock-in amps also measure phase.) If it's not the same, the output of the lock-in amp will be zero. See this description:

ِDear Amir 

No I did not find the answer for NIST 2004

I have not NIST 2008, If you have it, we may have beneficial negotiations.

Please let me know of your decision. 

Dear Gokul,

The basic parameter for ofdm system is the symbol time Tsy which is related to sub carrier frequency spacing deltaf. This time interval contains a number of samples equal to the total number of fft points comprising all the sub carriers whether it has a value or is nulled.

Therefore for 256 subcarrires one has 256 samples. In the frequency domain they cover a bandwidth= N deltaf, and in time domain then the sampling time will be Tsa = Tsy/N = 1/ N deltaf and the sampling frequency will be fs= 1/Tsa= N deltaf. The equality of the number of samples and the number of sub carriers is dictated by the equivalence of signal in the discrete time and  frequency domain.i 

wish you success

Maximiliano,

''Why the humans created the music?''  WE did'nt, only add some more. Many animal species have created musics much before our primate ancestors.  Why did biological evolution created it?  It makes senses that some early animals had to detect other moving animals through the sound they emit while moving and walking.  These type of sound contains a lot of rythms associated with moving animal bodies. The purpose of the detection is also to infer the type of movement, the type of animal doing it, the position and direction of trajectory, etc.  HOw this could be best achieved?  Through interpretation of the sound vibrations in the ears through the motor systems equipped for enacting such movement in the animal.  So I assume that the origin of music is this interpretation of sound through the self-enaction of the motor system.  At some point in evolution it became possible for some species to go futher and to actually produce sound using the motor system in ways that were pleasing for the animals and could be used for many social functions such as mate selection through musical selection or coordination of motions into a kind of swimming dance as for sea mammals, etc, etc.  

Music is a universal cultural elements of homo sapiens and music has many common elements with human languages.  Babies responded first to the musical part of the language sound and it is likely that humans became humans through a form of collective singing dancing. There are many empirical evidences for this hypothesis.  

It may be useful

Mr. Kishore,

The threshold setting depends on the spectrum sensing technique to be used.

The first thing you must do is to estimate the signal power when the channel is idle (white space). Assuming that the noise is zero mean, and follows a Gaussian distribution, the sensing threshold (in case of the energy detection) depends on the false alarm probability (Pfa) you need by means of the Marcum-Q function (or the Gamma cumulative distribution function). The desired Pfa determines the appropriate threshold value, according to some noise variance (noise power).

Important: if your application is real-time, the sensing time to use is an important parameter. You must select a number of samples to do signal processing, without decreasing so much the overall throughput of your system. This is a fundamental tradeoff.

I recommend to take a look to the paper attached for more details. It's a performance comparison between spectrum sensing techniques. The corresponding equations for threshold setting are given for different techniques.

You can specially use equation 10, where \gamma is the threshold value and N is the number of samples (sensing time). Note that the test statistic used is the normalized energy (see equation 9), where you need to estimate the noise variance (noise power in white spaces).

Note that in equation 10, the threshold value must be cumputed from the INVERSE Gamma cumulative distribution function. In MatLab this can be done using the funtion "gaminv(P,A,B)", where P is the Pfa, A = N/2 and B = 2/N.

I hope this helps.

Best regards,

Luis M Gato.

For me the coexistence of different wireless devices is the presence of one in the activity region of the other. It has a sense of neighborhood. From the conceptual point of view coexistent is possible  when they do not interfere with each other. So, every receiver must not receive an interference greater than its threshold value in its in and out of band.  So, the other transmitter must work at a power , distance, frequency, time and code such that its recieved power at the intended reciever must be smaller than the threshold.  If the direction is also known it can be used to control the interference between a foreign transmitter and intended receiver.This is the space division.

So there are the known division methods to prevent interference: Frequency, time, code division, space and power level control to control the interfere at certain receiver.

Best wishes 

Dear Dr. Mohan Chandra , thanks lot for the valuable information.

Many different options here.  How much noise is there?  Is the noise in-band or out-of-band?  How fast does the frequency of the 3-6 Hz vibration change?  How fast does the 0.33 Hz envelope change - both in amplitude and in frequency.

The following should work with either analog or digital signal processing:

Band pass filter the vibration signal to reject DC and low frequencies below 3 Hz and frequencies higher than 6 Hz.

Extract the amplitude of the vibration envelope.  Options include precision rectification followed by a low pass filter or peak-follower circuit with appropriate (e.g. ~0.5 s) decay time.

Multiply the envelope signal by a 0.33 Hz cosine wave and by a 0.33 Hz sine wave, and low pass filter each product to derive separate in-phase and quadrature signals.  The time constant of the filter determines the detection bandwidth.  Square and add the in-phase and quadrature signals and take the square root of the result to extract the amplitude of the 0.33 Hz envelope modulation.  For analog processing, the multiplication, square and square root operations can be implemented using 4-quadrant multiplier integrated circuits.

Alternatively, if the frequency and phase of the 0.33 Hz changes reasonably slowly, apply a phase-locked loop to extract a stable 0.33 Hz reference signal, and use this for synchronous (lock-in) detection of the 0.33 Hz envelope signal.  This can offer lower noise than the RMS detection scheme outlined in the preceding paragraph.

Dear Maged,

The effect of FOG on the transmission coefficient  of the electromagnetic waves is measured and the data are given in the Link:https://www.rand.org/content/dam/rand/pubs/reports/2006/R1694.pdf

These data  can be used to calculate the model parameters of the wireless channel.

In case of no multi path the data can be used directly to calculate the path loss. In case of multipath the impulse response of the channel will be sum of impulses randomly delayed from the first received impulse. The muyipath delay model depends on the terrain of the propagation environment and you may adapt the patern suiting your environment. I think that the fog will affect the strength of the received copies from the different paths.Therefore you can adapt a model similar to yours multipath one and modify the attenuation values of its different copies to suit the FOG propagation medium.

This is just a proposal to proceed in the solution of the problem.

Best wishes

Usually, if light is used in communications, it is not used for processing. It's used for the transmission medium. Processing is done in silicon.

The advantage of light in the transmission medium is not propagation speed, but rather bandwidth. So your Shannon capacity of the medium will typically be larger than RF media. Greater capacity is often referred to as "faster," because it can transfer more bits/second.

Any form of encryption or coding should also be applicable for use in light transmissions, I would think. Ultimately, if you use laser instead of incoherent light, you should also be able to use RF-like modulation schemes. Otherwise, with incoherent light, you will typically be limited to using "intensity modulation," and what they call wavelength division multiplexing (which is frequency division multiplexing).

Okay, so what do you mean by "processing speed will be faster"? Are you considering that the processors are also optical? So that their theoretically broader bandwidth will increase processing speeds? Perhaps. Here's a start, to see what some issues are and what some arguments for and against might be.

Nakagami channels are used when the received signal has contributions from both diffuse and specular scattering, i.e., the electric field is the sum of a strong component (which is not necessarily line of sight) and several contributions with less amplitude. The m parameter relates the amplitudes of strong and weak components. Rayleigh fading is obtained when m=1. The Nakagami model is in general very similar to a Rician channel, but its pdf has a closed form expression which is simpler to evaluate numerically (it does not contain Bessel functions) and fits better some measurements. There is also an equivalence between the K parameter in the Rician distribution and the m parameter in Nakagami. For a very brief introduction to multipath channels you can see Andrea Goldsmith's book:

Or if you need something more especific about Nakagami fading, you can check this physical model:

Why is Nakagami-m fading channel a good in practice in place of Rayleigh fading ? - ResearchGate. Available from: https://www.researchgate.net/post/Why_is_Nakagami-m_fading_channel_a_good_in_practice_in_place_of_Rayleigh_fading [accessed Jan 5, 2016].

Use fork in Tools box.

You can convert from Cartesian to Polar coordinates by using the CORDIC algorithm.

It is widely used in digital environments as an efficient tool for such conversation process. For detailed information please revert to the link:http://www.mathworks.com/help/fixedpoint/ug/convert-cartesian-to-polar-using-cordic-vectoring-kernel.html

wish you the best.

Yes, provided T2 represents the total information transmission duration over the entire period T.

Calculating LLRs requires generating a set of vector symbol candidates. Once you have a set of candidates, the LLRs can be computed the same way as a sphere decoder does it. How to generate such set? As an starting point, I'd suggest you to try the following approach:

1. Calculated the the ZF/MMSE equalized vector symbol.

2. Before performing hard decision, find the K closest constellation vectors to  your   ZF/MMSE equalized vector symbol.

3. Then you will have K+1 candidates from which you can compute the LLRs. (K obviously depends on the size of the constellation set).

Hope this helps.

Python (with scipy/numpy packages) s by far your best bet.

Hello,

Assume you would like to calculate the normalised cross correlation of two sequences, x(n) and y(n), of length N. Then

Normalised_CrossCorr = 1/N * sum{ [x(n) - mean(x)]* [y(n) - mean(y)] }/ (sqrt(var(x)*var(y))

where

>the sum is taken from 1 till N.

> mean(x) is the mean of x

>var(x) is the variance of x.

>sqrt is square root

I think you are talking about the minimum euclidean distance receiver, the two receivers you mentioned the Correlation receiver and the matched filter are 2 implementations of the minimum euclidean distance receiver. sometimes they are also known as ML receiver.  The output of the two implementations will be the same at a particular sample say t=(n+1)Ts. If only the value at the sampling instant is of

interest, then the matched filter and correlator. give the same result, as has been already noted. But, intermediate results are quite different .

If you want to design the LCL filter for a shunt-APF, the attached paper would be of help.

Good luck

I'm assuming by correlation, you mean the Pearson coefficient, http://en.wikipedia.org/wiki/Pearson_product-moment_correlation_coefficient

If that's the case, what you're assembling is a series of dot products between your new sample vector, and your collection of old sample vectors. Say you have new vector v, and labelled voice data vectors (x_i, y_i). If you think about the dual form of the training optimization procedure, a linear classifier -- the logistic regression, or the support vector machine -- is going to also be making its decisions based on a weighted average of  \sum_i \alpha_i x_i^Tv, where the precise values of \alpha will be set by the classifier's training algorithm as a function of how much it thinks the dot product of v with x_i is in terms of making new predictions. (Maybe your seventh recording of "HELLO" is garbled somehow; a good linear classifier can effectively learn to ignore that example.)

Your correlation approach is basically one option that a linear classifier (with an intercept term) is going to consider when it's performing its training procedure. Take a look at MATLAB's glmfit: http://www.mathworks.com/help/stats/glmfit.html It can help you produce these kinds of linear classifiers.

The main factor determining the impact of a channel (e.g. time-dispersive / doubly dispersive / optical channel) to an FBMC signal is the prototype filter of the FBMC Tx itself. Based on the Ambiguity Function of the filter, you can calculate the interference pattern in the time frequency lattice (TFL) after the signal passed through the channel. A good basic overview on these descriptions is e.g. given in the thesis of Jinfeng Du (http://www.ict.kth.se/publications/publications/2008/TRITA-ICT-COS-0803.pdf).

When we talk about receiver processing, there are several types of equalization algorithms and implementations, whose performance strongly interacts with other parts of your system model. Can you give more details about the channel model (dispersion, frequency-dependency etc.) of the channel you are considering?

Please see the attached pdf  which is based on earlier pdf . Hope it is useful.

Cheers

I think you can follow this method:

Suppose size(B)<size(A)

1. c=xcorr(A,B);     this gives you crosscorrelation between A and B (here B is zero padded to reach the size of A)

2. [YMax,XMax]=max(c);   this gives you the delay (Xmax) for which there is a maximum of correlation (Ymax) between A and B.

3. You see the parts of the signals A and B which are involved in the calculation of  (Ymax) and you take off all the other samples (for A and B).

4. Now you have two new signals A_hat and B_hat with size<=size(B); you apply the corr2 function and you have your correlation coefficient. 

Regard

Mourad

Nalin, The idea behind compressed sensing means we are sensing at compressed rate (below Nyquist rate). It is different from "sensing and compress" like in jpeg and others.  If one needs 20 sensors in a traditional sensing to sense a signal x of length 20. Then if x is sparse having, say 2 non-zero elements whose locations are unknown, then only 3 sensors (2*long20) will suffice instead. Further, if you know the location of those non-zero elements (2 in this example), then only 2 sensors will do the job.

On the basis of required accuracy of the channel estimation, optimal training sequences of minimum length are determined and it is given by

NP ≥ NT (L + 1) + L

Where, Where, Np = the number of training symbols per transmit antenna and per frame.

The training sequences very much effect the channel capacity, and by implementing the Optimal training sequence length improved MSE and capacity both.

Interesting question. The two are targeting different use scenarios, though. IEEE 802.22 is to provide wireless broadband over a metro area (so-called "regional area nets"), whereas IEEE 802.11 nets are local, shared LAN hotspots, hopefully much shorter range.

So, my thinking is, to the extent that any of these white space schemes can be shown to work well, there's no reason to believe that 802.11af and 802.22 cannot coexist peacefully. I would think that you can pack more 802.11af nets than 802.22 nets over the same RF channel, in a given area, because co-channel interference will be less pronounced with the lower power WiFi than with the RANs. In either case, it will be a combination of spectrum sensing and geo-location database that will determine whether or not the space is "white." And the presence of an 802.22 RAN will be felt much further away, for other systems wanting to use that channel.

To me, 802.22 is more of a tall stick, long(ish) range, TV-like standard than 802.11af.

Sorry I forgot to answer the question about channel estimation. The short answer is that it is the most practical to remove the dopper shift before processing the received signal.

As you are aware, a well-designed testing signal (training signal) must be sent over the channel and its signal properly recieved and analyzed at the receiver. The receiver must have an exact replica of the initial transmitted signal. It is equivalent to measuring the impulse respone of the channel although we often use the Fourier Transform equivalent and do the analysis in the frequency domain, i. e., use the transfer function. Doppler shift of the transmitted signal affects all parts of the signal spectrum but if we send a signal that has a carrier component, we can correct for doppler shift by estimating the received carrier frequency as compared to the local receiver clock.

In the time domain the signal repeates itself along time. In spatial domain, it repeats itself with position (say in the 'x' direction). In mathematical terms, just replace t with x. The period will not be in seconds but in metres. The frequency will not be in Hz (1/s) but in 'per metre', 1/m. Simple!

Could you please explain more why?

Hi,

You can always run any Simulink file from your .m file (or your optimization algorithm written in .m file) by simply using the following command:

simopt=simset('solver','"here you should define the solver you want to use','SrcWorkspace','base'); %if there is any variable needs to be read from workspace

sim('The simulink file name goes here',[0 0.1],simopt); %[0 0.1] is the simulation time which you can change it.

 

If you want to update the optimised parameter in Simulink at every step (let's say we're optimising a gain "kp" in Simulink file), you can add the following command to your .m file so that every time updates the parameter kp (which is "Gain" in this case with the current optimised value of kp in Simulink)

set_param('The simulink file name goes here/kp','Gain',kp);

 

I'm not sure if this is what you are after. Hope it helps.

I will try to answer this question in a very non-scientific manner.

Imagine a spiral whisk (see enclosed picture). If you keep the spiral part perpendicularly to your eyes and you look through, ignoring the 3D aspect of the object, you will see a modulated sine signal (a 2D projection), lets call it x(t).

The Hilbert transform shifts the 2D projected signal x(t) by 90 degrees in phase to create a signal y(t) = hilbert[x(t)] so that if you combine the original front projection x(t) and its Hilbert transform y(t) you got something we call the analytic signal z(t) = x(t) + j y(t). The analytical signal is, in our case, exactly the spiral whisk.

In terms of mathematics, the front projection of the whisk x(t) is the real part, the top projection y(t) would be the imaginary part, and the whole 3D spiral whisk is the analytic signal z(t), so z(t) = x(t) + j y(t).

Now, obviously, when you take the analytic signal (the spiral whisk), it is easy to extract the envelope part (the absolute value) and the carrier (phase, that can be modulated).

Dear Zeynab,

The multipath wireless channel is characterized by a correlation bandwidth Bc which is by definition the the inverse of the delay spread time. That is Bc= 1/( Tn-1 - To), where Tn-1 is the largest delay and T0 i s the smallest delay of the received signal at the receiver. If the signal transmitted across this channel is smaller than Bc, then you shall observe flat fading. Frequency selective fading will be observed when the transmitted signal bandwidth is greater than Bc. Then i would propose to model the transmitter and the receiver in matlab/ simulink and connect a Rayleigh fading channel. Set Bc of the channel and vary the bandwidth of the transmitted signal by decreasing the symbol time. When you realize the above conditions you will get flat fading or selective fading according to the above condition.

Thank you for your interesting question.

wish you success.

For more detail please follow the link:

I know that the sampling rate can influence the quantization noise. Its spectrum can be strongly correlated to the input signal (e.g., low amplitude periodic signals). The correlation depends on the ratio between the sampling frequency and the signal frequency. Solutions: fine-tuning the sampling frequency; summing up the signal with a small amount of white noise.

You can check agilent site for more details

The core of fibres is very tiny. Therefore, increasing the power may increase the power density in the core to a level where the material shows non-linear effects, such as Brillouin scattering. This, in turn, brings up unwanted frequencies to propagate in the fibre, with the consequence of unwanted cross-modulation terms in the photodiode in the receiver.

The step size should satisfy the condition:

0 < step-size < 1/(max eigen value of the input autocorrection matrix)

For fast convergence step-size is set close to its maximum allowed value.

I mean, in every iteration of the code, the frequencies that I should filter them are different. So, I need a function of frequency (like a wavelet) that I can multiply it into the signal in every iteration.

As per your second question: I assume that, you are asking about EMBEDDED processors. A 16b processor (e.g., dsPIC series from Microchip) can process data samples 16b at a time. A 32b processor can process them 32b at a time (e.g., PIC32MX series from Microchip).

However, don't forget, most 8 bit processors have a 16bit data type, and they use two registers to describe a 16 bit number. For example, if you use a 12 bit ADC as an input to a 8 bit processor, this is perfectly fine. But, you will be using more MCU instructions to process it, since it will map to TWO natural data types (i.e., byte) inside.

In the website from Qualcomm you can find two application examples of Raptor Q codes , which are a special type of fountain codes:

Concretely, the applications are delivery of digital cinema movies over satellite to a many of cinemas and a solution for videoconferencing.

Raptor codes are also part of the MBMS standard and a proposed standard from IETF (RFC 5053, RFC 6330).

At DLR, where I am working, we have developed a working implementation of a fountain encoder/decoder, and we have demonstrated them in the lab. Currently there are plans to launch a satellite which can perform onboard fountain encoding to deliver files to several satellite terminals. For more information see:

There are a lot of issues to be considered in multicarrier systems, ranging from physical layer (e.g., transceiver design, channel estimation) up to upper layers (e.g., resource allocation, carrier balancing, power allocation, interference management).

You can find an adequate overview of these topics in these papers.

The evolutionto 4G cellular systems: LTE-Advanced