7CCEMCTH COMMUNICATION THEORY
Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit
COMMUNICATION THEORY
7CCEMCTH
COURSEWORK
There are 4 Questions, answer all.
Detailed answers and calculations are required.
You may use MATLAB where required.
You may use the DFE program, only where you are asked by the questions.
Upload clearly scanned pdf copies of your written answers by the deadline, as indicated on Keats.
Question one (Finite-length Design) [25 marks]
Figure 1: Finite-length transmission system design
The sampled overall pulse response, i.e., p(t) = p(t) ∗ ℎ(t) ∗ ℎaa (t) , of a 2-ray transmission system in Figure 1, is given by
pk = 6k + 0.7j6k - 1,
The oversampling factor is unity, i.e., l = 1 , the sampling has been done at symbol rate. Quadrature amplitude modulation (QAM) symbols are transmitted with average energy per dimension of Ex = 1 . The noise variance is a 2 = = 0. 149 .
Hint: Follow direct formulations for answering the following parts of this question. Although, solutions using the DFE program are not acceptable and will not be marked for Question one, you can use it to double check your solutions for yourself.
a) By following the finite-length formulations, design a minimum mean square error-linear equaliser (MMSE-LE) filter, by calculating the discrete-time impulse response wk with 3 coefficients, i.e., Nf = 3 , and the resulting signal to noise ratio (SNR), i.e., SNR MMSE -LE,U , with an overall delay factor of Δ = 0 . (7 marks)
b) Repeat part (a) with for a more realistic overall delay factor of Δ = 2 . (4 marks)
c) By revisiting the analytical formulations, determine/estimate what delay coefficient would result in a best SNR MMSE -LE,U for this channel, although it
may not always be a practicl one . (2 marks)
d) Repeat part (a), for a minimum mean square error-decision feedback equaliser (MMSE-DFE) with Nf = 2 and one feedback tap, i.e., Nb = 1 , and find the feedforward and the feedback filters, coefficients as well as the achievable SNR MMSE -DFE,U by directly following the finite-length formulations
with an overall delay factor of Δ = 1 . (10 marks)
e) Compare your results in parts (a) and (d) in terms of complexity of implementation and the achievable SNR after equalisation using MMSE-LE and MMSE-DFE and conclude. (2 mark)
Question two (Finite-length Design) [25 marks]
Figure 2: Finite-length transmission system design
For the transmission system in Figure 2, the two-ray channel impulse response in the baseband is given as
ℎ(t) = 6(t) + 0.7j6(t 一 T).
The impulse responses of the transmit filter, i.e., the basis function p(t), and the anti-aliasing filter, i.e., ℎaa (t), are:
p(t) = sinc ( ),
l lt
where l is the oversampling factor and T is the symbol period. Quadrature amplitude modulation (QAM) symbols are transmitted with an oversampling factor of l = 2 and an average energy per dimension of Ex = 1 . The noise variance is o2 = = 0.149.
a) Find the expression for the overall channel pulse response p(t) for the finite- length design. (2 marks)
b) Find the discrete-time expression for the oversampled p(t), i.e., pk , with an oversampling factor of l = 2. (2 marks)
c) Calculate the oversampled coefficients of p(t) with l = 2, keeping an overall duration of 5T of p(t), i.e., when 一2.5T 三 t 三 2.5T, and ignoring the rest of the energy of p(t). (3 marks)
d) Express the resulting discrete-time pk signal in part (c) in D-transform domain. (1 marks)
e) Find the causal expression for the pk samples in discrete-time domain, i.e., the oversampled channel taps in part (d). (2 marks)
f) Find the matrix expression for the causal pk samples in part (e), i.e., the expression [p0 p1 ⋯ pv], where pi is a l 人 1 vector of respective pk coefficients. (3 marks)
g) Using the DFE program, design a minimum mean square error-linear equaliser (MMSE-LE) digital filter (with fractionally-spaced delay line) by finding its discrete-time impulse response coefficients wk and the resulting unbiased signal to noise ratio (SNR), i.e., SNR MMSE -LE,U , with Nf = 5 (i.e., number of coefficients corresponding to the symbol spaced delay line of the MMSE-LE filter). Let the DFE program chooses the best delay Δ , i.e., set Δ = 一1 in the program. (2 marks)
h) By experimenting with the DFE program in part (g), find the Nf value that you would choose for your MMSE-LE design as an efficient compromise between the cost and the performance, e.g., a higher achievable data rate?
Hint: Do not write the filter coefficients. Only, write down the largest achievable SNR MMSE -LE,U , the corresponding Nf value and the best delay chosen by the program. (2 marks)
i) Repeat part (g) for Nf = 3 and record the corresponding achievable SNR MMSE -LE,U , filter coefficients and the best delay chosen by the program. (2 marks)
j) Repeat part (g) for a minimum mean square error-decision feedback equaliser (MMSE-DFE) with Nf = 3 and one feedback tap, i.e., Nb = 1 and record the corresponding feedforward and feedback filters, coefficients as well as the achievable SNR MMSE -DFE,U and the corresponding best delay chosen by the program. (2 marks)
k) By experimenting with the DFE program, what combination of number of feedforward and feedback taps, i.e., Nf and Nb , would you choose as a system designer for this data transmission system to obtain an efficient compromise between the complexity cost of practical implementation and the performance in terms of SNR MMSE -DFE,U .
Hint: Do not write the filter coefficients. Only, write down the largest achievable SNR MMSE -LE,U , the corresponding Nf and Nb values and the best
delay chosen by the program. (2 marks)
l) Compare your results in parts (g) and (j) in terms of the complexity of implementation and the achievable SNR after equalisation using MMSE-LE and MMSE-DFE and conclude. (2 marks)
Question three (Infinite-length Design) [25 marks]
In a filtered additive white Gaussian noise (AWGN) channel a ratio of = 25 in linear scale is assumed, where Ex and G2 are the average energy of transmission per dimension and the variance of noise, respectively. The target probability of symbol error on this channel is pe = 10一6 , i.e., the gap is Γ = 8.8 dB. The channel discrete- time deterministic autocorrelation function in D-transform domain is given as
Q(D) + = 0.5(jD一1 + 2.04 一jD),
where SNRMFB is the matched filter bound on signal to noise ratio (SNR).
a) Writing the details of your calculations, find the SNRMFB . (3 marks)
Assuming SNRMFB = 50, answer the following questions:
b) Design a minimum mean square error-linear equaliser (MMSE-LE) for a
transmission system on this channel by finding its filter transform
WMMSE -LE (D) . (3 marks)
c) Compute the achievable unbiased SNR, i.e., SNR MMSE -LE,U , for your design in part (b). (5 marks)
d) Design a minimum mean square error-decision feedback equaliser (MMSE-
DFE) for a transmission system on this channel by finding its feed -forward
and feedback filter transforms, i.e., WMMSE -DFE (D) and BMMSE -DFE (D) . (6 marks)
e) Compute the achievable unbiased SNR, i.e., SNR MMSE -DFE,U , for your design in part (d). (2 marks)
f) Successive quadrature amplitude modulation (QAM) symbols with average power of P = = 20dBm, where Ex is the average transmit energy and T is the symbol period, are transmitted through this channel. The power spectral density of noise is G2 = 一70 dBm/Hz. (Hint: 0dBm= 1mW.)
i. Which one of your designs in parts (b) and (d) results in a higher maximum data rate on this channel? Why? (1 mark)
ii. Find this higher maximum data rate in part (f-i) with an integer number of bits per symbol. (5 mark)
Question four (Channel Characterisation) [25 marks]
The magnitude square of the frequency response of a filtered additive white
Gaussian noise (AWGN) channel is shown in the baseband in Figure 3.
Figure 3. Channel,s magnitude-squared frequency response
Successive quadrature amplitude modulation (QAM) symbols are transmitted through this channel with an average transmit power of = 80 mW, where T is the symbol period and Ex is average energy of transmission. The power spectral density of the additive white Gaussian noise is = 一91 dBm/Hz. The basis function of transmission is p(t) = sinc (). The sampling rate is at symbol rate and the target probability of symbol error is set at Pe = 10-6 (i.e., with a gap of 8.8 dB).
a) Find the matched filter bound on signal to noise ratio (SNR), i.e., SNRMFB , at a symbol rate of = 20 MHz for this transmission system. (6 marks)
b) At the symbol rate of 20 MHz, determine the unbiased SNR for a minimum mean square error-linear equaliser (MMSE-LE) receiver that may be used in this transmission system. (3 marks)
c) Using the gap formula, find the maximum achievable data rate in part (a), with an integer number of bits per symbol, and identify the size of required QAM constellation. (3 marks)
d) Find the matched filter bound on the SNR, i.e., SNRMFB , at a symbol rate of = 40 MHz for this transmission system. (7 marks)
e) Using the gap formula, find the maximum achievable data rate in part (d), with an integer number of bits per symbol. (2 marks)
f) Can the data rate in part (e) be approached in practice without channel equalisation? Why? (2 marks)
g) Compare your results in terms of maximum achievable data rates at symbol rates of 20 MHz and 40 MHz and explain your reason for a possible difference in terms of achievable number of bits per symbol, symbol rate and the channel spectral characteristics. (2 marks)
2023-03-17