MOB sujet 4 Transmission techniques - DS CDMA

MOB sujet 4 Transmission techniques - DS CDMA

MOB sujet 4 Transmission techniques - DS CDMA

1 CDMA Exercises

The principle of the spread spectrum modulation consists in coding the information by the way

of a bandwidth signal larger than the band strictly necessary, with a goal of fighting the interfering signals and the distortions tied with the propagation

The signal is coded at the beginning: a code is assigned with each user in order to allow decoding in the arrival The spread is ensured by a pseudo-random signal called spread code At the reception, the signal is seen as the noise if the receiver does not have the code

The spread spectrum modulation is optimized to fight against the noise, by which it limit better the effects

1.0.1 Mathematical reminds

- Correlation measure the manner whose two signals vary simultaneously A strong

correlation implies a great resemblance between the two compared signals

- Auto-correction quantifies the correlation of a signal with a version of itself shifted in

time, according to this temporal shift

- Cross-correlation quantifies the correlation of two shifted signals the one in comparing

with the other in time, according to this temporal shift

- Orthogonality: two periodic signals from period T are known as orthogonal if their

cross-correlation is null for a null temporal shift

CDMA modulation (Code Division Multiple Access) is a spread spectrum modulation It allows transmitting two signals simultaneously in the same band, their separation being realized at the reception in getting profit of their orthogonality Several mobile telephony systems (IS95, UMTS…) have chooses this technique The CDMA is then also used in order to manage the multi-users aspect Among the spread spectrum methods are:

- Frequency Hopping - CDMA (FH-CDMA), seemed to be the most interesting technique

at the beginning, but its difficult implementation took the interest back on spread spectrum

by directed sequence;

- Direct Sequence - CDMA (DS-CDMA), is used in particular in the 3G European UMTS

system (Mobile Universal Telephone System)

The principle of the DSCDMA is as follows On each transmitting station, is allocated a

single code of m chips called "chip sequence" constituted from "-1" and from "+1" To

transmit a bit to "1", the station sends its sequence chip To transmit a bit to “0”, the station

sends the opposite of its sequence chip The duration Tc of transmission of a chip is m time lower than the duration Tb of a bit.

Example: let us suppose that the chip sequence of A is "-1-11-1111-1 " (m = 8) To transmit

a binary "1", A sends "-1-11-1111-1"; to transmit a binary "0", A sends "11-11-1-1-11" Power spectral density (PSD) of a signal corresponds to the power distribution of the signal according to the frequency For a code of forming, the PSD is proportional to the square of the Fourier transform of this forming The forming is a form of wave given starting from

which one announces a "0" or a "1" If this is about one rectangular function of width T, the Fourier transform of such a function is a cardinal sine whose first cancellation is of f = 1/T.

One thus finds a PSD which has the form of the square of a cardinal sine:

A signal DS-CDMA corresponds to a succession of rectangles of width Tc, the duration of

a chip Its PSD thus has the form of the square of a cardinal sine in the first cancellation by

f = 1/Tc.

By opposition, a digital signal is constituted by a succession of rectangles of width Tb, the

bit duration, has a PSD of the form of the square of a cardinal sine with the first

cancellation of f = 1/Tb.

The chip sequence consists of m chips Each bit of the digital signal in starting is replaced

by the chip code so that Tb = mTc And thus,

T

m

T C b



1

The principal lobe of DS-CDMA signal is thus m time broader than the principal lobe of the starting signal It is said that the signal was spread with a spread factor of m Two

signals have the same total power, but power of the starting signal is more concentrated in the band, where the appellation of “signal with narrow-band "

1.0.2 Exercises

1 In your opinion, what are the advantages and disadvantages in the use of the CDMA for the cellular networks?

Advantages: The spread spectrum modulation is optimized to fight against the noise, by which it limit better the effects

Disadvantages: In CDMA, any speaker can talk at any time; however each uses a different language Each listener can only understand the language of their partner As more and

more couples talk, the background noise (representing the noise floor) gets louder, but

because of the difference in languages, conversations do not mix

2 If one wishes to send information of b bits/s with a key of m chips, what will be the flow in

chips per second?

The chip sequence is constituted of m chips Each bit digital signal of departure is replaced

by the code chip so that Tb = mTc So, b

b c

T

m T

f  1   .

According to Shannon’s theorem, we deduce: Cc = mCb = mb chips/s

3 Which factor will the frequency band occupied by the signal be increased with?

The data bit rate and noise power ratio

The main lobe of DS-CDMA signal is m times larger than the main lobe of the signal of departure The frequency band occupied is increased by a factor m They say that the signal was unfurled with a spreading factor of m Both signals have the same overall power, but the signal of departure is more concentrated in the band, hence the name "narrow band signal”

4 Two keys are orthogonal if the standardized internal product of these both keys is null

Remind: internal product of [a, b][a’, b’] = aa’ + bb’.

Show that if two keys S and T are orthogonal, S and T are too.

Suppose S = [s1, s2, ,sm], T = [t1, t2,…,tm]

Thus T = [ t1,t2,…,tm]

We have S T = (s1 t1+ s2 t2 + … + sm tm)/m = 0

So S T = (s 1 t1 + s2 t2 + … + sm tm)/m =(s1t1+ s2 t2 + … + sm tm)/m = 0

e.g S and T are orthogonal.

5 One considers 4 transmitters having each one a key:

A: (-1 -1 -1 +1 +1 -1 +1 +1)

B: (-1 -1 +1 -1 +1 +1 +1 -1)

C: (-1 +1 -1 +1 +1 +1 -1 -1)

D: (-1 +1 -1 -1 -1 -1 +1 -1)

a) Show that B, C and D are orthogonal at A

A B = [(-1)(-1) + (-1)(-1) + (-1) (+1) + (+1)(-1) + (+1)(+1) + (-1) (+1) + (+1)

(+1) + (-1)(-1)]/8 = (1 + 1 – 1 – 1 + 1 – 1 + 1 – 1)/8 = 0

b) A code has a good autocorrelation if its product intern with itself is high and its internal product with itself is shifted of 1 or several chips are weak That means that the receiver will be able to recognize easily the emitted signal Show that the key of Barker (+1,-1,+1,+1, -1, +1,+1,+1, -1,-1,-1) is a good autocorrelation

We have: BkBk = 1

The value is even more important that the key is long, but it takes to get complete the concept of autocorrelation view the product's internal key with it even when it is offset by 1 chip, 2 chips,

Suppose Bk’ is Bk shifted of 1 chip to the left, thus:

Bk’ = (-1,+1,+1, -1, +1,+1,+1, -1,-1,-1,+1)

We have: Bk Bk’ = [1 x (-1) + (-1) x 1 + 1 x 1 + 1 x (-1) + (-1) x 1 + 1 x 1 + 1 x 1 + 1 x (-1) + (-1) x (-1) + (-1) x (-1) + (-1) x 1]/11 = [(-1) + (-1) + 1 + (-1) + (-1) + 1 + 1 + (-1) + 1 + 1 + 1]/11 = –1/11

The product of the key Barker (+1 -1 +1 +1 -1 +1 +1 +1 -1 -1 -1) with the same key but shifted by two chips to the left, Bk” (+1 +1 -1 +1 +1 +1 -1 -1 -1 +1 -1) is:

Bk  Bk” = (1 – 1 – 1 + 1 – 1 + 1 – 1 – 1 + 1 – 1 + 1)/11 = –1/11

We deduce that this key and the same key shifted by one chip or two chips are nearly orthogonal The higher the value of domestic product, the lower the receiver can easily

interpret the result (signal codes with keys "pseudo-orthogonal").

c) When a station wants to transmit a binary "1", it sends its key and when it wants to transmit a binary "0", it sends the opposite of this key If only C wishes to send a bit of

"1", what the receiver will receive and will make to interpret the answer?

If only C wishes to send a bit of "1", C send chip sequence S = (-1 +1 -1 +1 +1 +1 -1 -1) The receiver will calculate, it had: S A = 0 and S B = 0 and S D = 0 and S C = 1, so receiver know C was sent bit "1"

d) If only C wishes to send a bit of "0", what the receiver will receive and will make to interpret the answer?

If C send a binary "0", C send chip sequence S = (+1 -1 +1 -1 -1 -1 +1 +1)

The receiver will calculate, it had: S A = 0 and S B = 0 and S D = 0 and S C = –1, so receiver know C was sent bit "0"

e) The signals are added linearly, the transmitters are synchronous and the signals are received with the same power by the receiver If one supposes that the stations A, B and C decide to send at the same time the bit "0", what will be the sequence read by receiver?

The stations A, B and C decide to send at the same time the bit "0", chip sequence S at the receiver:

S = (+3 +1 +1 -1 -3 -1 -1 +1)

(Remember we have: A = (+1 +1 +1 -1 -1 +1 -1 -1) and B = (+1 +1 -1 +1 -1 -1 -1 +1) and

C = (+1 -1 +1 -1 -1 -1 +1 +1) )

f) If the receiver receives the sequence “-1 +1 -3 +1 -1 -3 +1 +1”, what does it deduce from?

Suppose S = (-1 +1 -3 +1 -1 -3 +1 +1)

We have:

S A = (-1 +1 -3 +1 -1 -3 +1 +1) (-1 -1 -1 +1 +1 -1 +1 +1) = (1 - 1 + 3 + 1 – 1 + 3 + 1 + 1)/8 = 1, so A sent bit “1”

S B = (-1 +1 -3 +1 -1 -3 +1 +1) (-1 -1 +1 -1 +1 +1 +1 -1) = (1 – 1 – 3 – 1 – 1 – 3 + 1 – 1)/8 = -1, so B sent bit “0”

S C = (-1 +1 -3 +1 -1 -3 +1 +1)(-1 +1 -1 +1 +1 +1 -1 -1) = (1 + 1 + 3 +1 – 1 – 3 – 1 – 1)/8

= 0, so C has not sent anything

S D = (-1 +1 -3 +1 -1 -3 +1 +1) (-1 +1 -1 -1 -1 -1 +1 -1) = (1 + 1 + 3 – 1 + 1 + 3 + 1 – 1)/8

= 1, so D sent bit “1”

6 Suppose that one wants to transmit a data flow of 56 kbit/s by using a spread spectrum technique

a) Find the band-width of the necessary channel when the signal-with-noise report is 0,1; 0,01 and 0,001

Case 1: SNR = 0,1

C = B log2(1 + SNR) B = C / log2(1 + SNR) = 56 103 / log2(1 + 0,1) = 407.262Hz

407,2KHz

Case 2: SNR = 0,01

C = B log2(1 + SNR) B = C / log2(1 + SNR) = 56 103 / log2(1 + 0,01) = 3.901.000Hz

3,9 MHz

Case 3: SNR = 0,001

C = B log2(1 + SNR) B = C / log2(1 + SNR) = 56 103 / log2(1 + 0,001) = 38.835.647Hz

 38,835 MHz

b) In a ordinary system (without spread spectrum)

In a regular system (without spread-spectrum), so that R = B Shannon's formula C = Blog2 (1 + S / N) applied with C = B = R gives log2 (1 + S / N) = 1 This implies that S / N = 1 Thus, spread-spectrum without a S / N much greater is required

7 Prove that a receiving station can get the data sent by a specific sender if it multiplies the entire data on the channel by the sender’s chip code and then divides it by the number of stations

Let us prove this for the first station, using our previous four-station example We can say that the data on the channel:

D = (d1 c1 + d2 c2 + d3 c3 + d4 c4)

The receiver which wants to get the data sent by station 1 multiplies these data by c1

D c1 = (d1 c1 + d2 c2 + d3 c3 + d4 c4) c1

= d1 c1 c1 + d2 c2 c1 + d3 c3 c1 + d4 c4 c1

= d1N + d20 + d30 + d4 0

 d1 = (D c1 ) / N

2 Practical - Spread spectrum techniques

The goal of this practical is to understand the impact of spread spectrum techniques on the emitted signal and on the received signal by the recipient You will be able with this practical: – to visualize and study impact of the code used and of the modulation type on the spectrum

of a signal;

– to observe the properties of two signals according to the used keys (DS-CDMA);

– to visualize quality of reception of a signal 1, while varying the signal with noise report (SNR) as well as the ratio (C/I) of the power of the signal has to interpret (Carrier: signal 1)

by compared with a second signal (Interferor: signal 2) which will come to disturb the reception of the first;

– to trace the curves of Bit Error Rate (BER), according to coding for a SNR or a fixed C/I

For that you will use the simulation tool developed for this effect Copy the script tme_cdma in

your folder:

Carry out the script with the help of Octave, the digital calculation, visualization and programming

software similar to Matlab Appear the following menu:

Choose an option

[1] signal key 1 (useful)

[2] signal key 2 (interfering)

[3] display the spectrum of the useful signal (red) and the interfering signal (green)

[4] choose the modulation type

[5] trace the autocorrelation of the key 1 and the cross-correlation between two keys

[6] trace the constellation diagram in reception of signal 1 (useful) for a SNR and a fixed C/I

[7] calculate the bit error rate in reception for a fixed SNR

[8] calculate the bit error rate in reception for a fixed C/I

[9] quit

Menu description

[1]: allows visualizing or modifying the spread spectrum key of the "signal 1" The key by defect is the key of Barker used in Wifi?

[2]: allows visualizing or modifying the spread spectrum key of the "signal 2" The key by defect

is a sequence of "1"

[3]: trace the curves of the spectral density of signals 1 and 2 after modulation with and without filtering The two traced spectra, in red for signal 1 and in green for signal 2 have both the same power This spectrum is obtained for each signal by getting the same continuation random of 10 000 bits Each signal is coded with its own key which can be visualized and modified while typing [1] or [2] in the menu

[4]: allows choosing the modulation type which will be applied to signal 1 You have the choice between: PSK-2 (BPSK), QAM-4 (QPSK), QAM-16 and QAM-64

[5]: allows visualizing the curve of the autocorrelation of the signal 1 key and cross-correlation between the keys of signals 1 and 2

[6]: allows visualizing in the form of points in a constellation diagram, the value of the different symbols received by the receiver One can thus see the amplitude and the phase of these symbols The configuration of the receiver is adapted according to the modulation type of signal 1, of the CDMA keys of the two signals, and the values of ratios SNR and C/I

[7]: allows tracing curves of BER according to report C/I for a fixed SNR?

[8]: allows tracing curves of BER according to report SNR for a fixed C/I?

3 Properties and spectral of the emitted signal

The principle of the DS-CDMA is as follows At each transmitting station is allocated a

single code of m chips called "sequence chip " constituted from "1 " and from "+1" From a

code constituted from "0" and from "1" like the code of Barker, it is advisable to replace the

"0" by "1" to obtain the sequence chip To transmit a bit to "1", the station sends the opposite of its sequence chip To transmit a bit to "0", the station sends the opposite of its

sequence chip The duration Tc of transmission of a chip is m time lower than the duration

Tb of a bit.

1 What is the key value by defect of the signal 2? What does it mean? For the first part of the tests, we keep the key value of the signal 1 on

"10110111000" and the key of the signal 2 on "1" There is thus nothing to modify

2 (a) Choose a modulation type by using the menu [4] and display then the spectral density of power of signal 1 and signal 2 by using the menu [3] What do you observe by comparing the form of the primary and secondary lobes of signals 1 and

2 without filtered for the selected modulation? These observations are they still valid for the other modulations?

(b) What is the power reduction between the primary and secondary lobe for the two curves?

(c) What is the effect of the filtering on the frequential spectrum?

(d) What is the interest of the filtering?

4 Properties and correlation

The autocorrelation is studied in considering the profile of the autocorrelation function

obtained by shifting the digital sequence x[k] (of length N completed beforehand by zeros) of a gradually increasing value n and by comparing the initial sequence and the shifted sequence.

The formula used here to estimate the autocorrelation function is written:

The correlation coefficient obtained will be larger as much as the digital sequence and its shifted version are similar Conversely, if the sequences are opposed, the correlation will be negative A null autocorrelation indicates that there is not correlation between the initial sequence and the shifted sequence for the considered shift

The study of the autocorrelation allows us to evaluate the capacity of the receiver has to synchronize with the flow of chips of the transmitter In effect, the receiver makes a correlation between the received signal and the used key in the emission A peak of correlation must be observed when on the duration of a bit, the key used for encoder a bit and the key used by the receiver are synchronous In the other hand, the value of γx[n] must be near to zero for a shift not null representing a desynchronization of the receiver Thus, a good key must have such properties It will be said that it has good properties of autocorrelation

1 A good autocorrelation must present a maximum for a null shift and must be near to zero elsewhere Which interest is presented in the case where the channel is the subject of multi-traffic?

2 Observe the autocorrelation of the Barker key of the IEEE standard 802.11

"10110111000" by using the menu [5] It appears good to you?

3 Which property appears important to you to choose the keys of two distinct transmitters?

By basing you on that which was known as the autocorrelation properties, what appears

important to you for the cross-correlation?

4 Give the mathematic expression of an estimator of the cross-correlation of two sequences x[k] and y[k]

5 Are the keys of signals 1 and 2 orthogonal?

6 Modify the keys of two transmitters and observe their autocorrelation and their cross-correlation

5 Reception and interpretation of the signal 1

The menu [6] makes it possible to visualize the value of the amplitude and of the phase of different symbols received by the receiver The simulator allows observing the impact of the modulation; of interferences linked to the additive noise (SNR) and to the disturbance caused by another signal (signal 2) synchronous with the first

1 What does it mean SNR or ratio C/I positive, null and negative?

2 Reposition the key value of the signal 1 to “10110111000" (Barker key) and that of signal 2 to “1”, i.e to the default values Using the menus [4] and [6], vary the SNR, the C/I and the modulation type and fill out table 1 What do you observe? what does it occur when ratio C/I tends towards 0?

Tab 1 – Synthesis of observations on the constellation diagrams

Modulation SNR C/I Quality of the reception of signal

(Good/ Medium /Bad)

3 Position the key value of the signal 1 to "10110111000" (Barker key) and that of signal 2

to "01101110001" Signal 1 and signal 2 are then encodes by orthogonal pseudo keys Tab 2 – Synthesis of observations on the constellation diagrams

Modulation SNR C/I Quality of the reception of signal

(Good/ Medium /Bad)

The signal 2 simulates here a signal interfering broad band Vary the SNR, the C/I and the modulation type and fill out table 2 What do you observe?

4 In basing itself on the data of the exercise 5 of part 1.3.2 of the subject of precede practical, one associates now the key "00011011" to signal 1 and the key B "00101110"

to signal 2 Fill out table 3 What does one observe then? Explain

Tab 3 – Synthesis of observations on the constellation diagrams

5 What does it occur if one associates the Barker key to the two signals?

a) Fill out the table 4

Tab 4 – Synthesis of observations on the constellation diagrams for codes of Barker

Modulation SNR C/I Quality of the reception of signal

(Good/ Medium /Bad)

b) What does one observe then? Explain

c) Why would the results on the constellation diagrams have the same ones without use of CDMA codes (key of the signals is "1")?

d) What would have been different without use of CDMA codes (key of signals is

"1")?

e) If one preserved the Barker key for the two signals, one could have had a result different from which is showed here by the simulator? Explain

Modulation SNR C/I Quality of the reception of signal

(Good/ Medium /Bad)

f) How can you highlight with help of the simulator?

6 Curve of BER

Trace a curve of BER in defining the type of modulation, the fixed size (SNR or C/I), the intervals of variations interesting for the variable ratio The calculation of the curve taking several minutes, it is advised to limit the number of points to be displayed Does one obtain the same thing if the modulation is changed, the fixed size?