e e ee 7` w sgiri/pdfs/ee4140-2020/ee4140-assg1...title assg1.jnt author giridhar created date...
TRANSCRIPT
Assignm
ent #1 -- Oct.21, 2020
1. [5 mark
s] We consider raised cosine (R
C) pulse shaped transm
ission. The im
pulse response of the pulse-shaping continuous time filter g(t) is given
as follows:
g(t) = S
inc( t/T) { C
os(t/T
) / (1-42t 2/T
2) }
where the excess bandw
idth factor 0 ≤≤
1 and T is the sym
bol duration and Sinc(x)=
Sin(x)/x. W
e use a discrete-time filter g(nT
s) = g(n) w
here the sam
pling rate Ts =
T/J, w
here J is a positive integer. Also, the filter is truncated to 2L
T sym
bol durations, i.e., LT
symboldurations on each side of t=
0, w
here L is a positive integer. T
he input bit (or symbol) stream
I(kT) into this filter w
ill be zero-interleaved with J zeros in order to create the output
sequence x(kTs) w
hich is then sent into a DA
C.
Chose your ow
n 16-bit random sequence w
here I(k){+
1, -1} in order to create a BP
SK
signal. Assum
e all the bits before and after this 16-bit pattern are zeros. Y
ou are to plot the output sequence (samples) x(kT
s) using both “symbol-plot” as w
ell as join them using “line-plot” using M
atlab.
(a) Plot as F
ig.1, x(kTs) for J=
4, L=
2, and =
0.(b) R
epeat (a) and include in Fig.1 another plot with the change L
=4.
(c) Plot as Fig.2, three
different plots of x(kTs) w
here all of them have J=
8, L=2, but w
ith threedifferent excess bandw
idth factors, namely,
=0,
=0.5,
and =
1.(d) C
omm
ent on your results in (b) and (c).
2. [15 mark
s]In this question, w
e compare the theoretical probability of sym
bol error P(e) =
PS
and/or probability of bit error, PB , w
ith computer sim
ulated sym
bol or bit error rate (SE
R or B
ER
), for a few popular linear m
odulation schemes. T
he approximation on P
Susing bounds w
ill also be studied.
(a) For the same energy per bit, E
b, compute and plot on the sam
e figure (Fig-1) the PS
of BP
SK
, QP
SK
, and 16-QA
M signals. A
ssume the com
plex base-band A
WG
N m
easurement m
odel r(k)=I(k)+
v(k), where iid I(k) is uncorrelated w
ith the white G
aussian noise v(k). For complex signals
(QP
SK
and 16-QA
M) the noise
will be com
plex with variance N
O /2 per dimension. V
ary the ratio 10log10 (E
b/No) in 2dB
steps from 0dB
to 10dB, by changing the noise variance and plot against
log10 (P
S ), for all the 3 signals. Chose E
b=1 for all the signals. N
ote that the log of the error probability is plotted on the y-axis. The Q
(.) function or the erfc(.) function can be evaluated using M
atlab.
(b) Assum
ing Gray coding, plot 10log
10 (Eb/N
o) versus log10 (P
B ) for the 3 signals. Add these curves also to F
ig-1, and label neatly.
(c) For the QP
SK
signal set, compute using the follow
ing approximations to the sym
bol error probability. Plot each of these values of P
S for the range of SN
Rs
from 0dB
to 16dB in the sam
e figure. Call it Fig-2.
(c1) Union bound using allthe pairw
ise symbol errors
(c2) Union bound using only the nearest neighbours
(c3) In part (c2) replace the erfc() function with the C
hernoff bound and add this also to Fig-2.
(d) Now
, we w
ish to simulate the sym
bol error rate (SE
R) for Q
PS
K. For generating the transm
it symbols, use uniform
random variable (rv) betw
een (0,1). If the rv takes a value X
between 0 and 0.5, it is m
apped to say -d; else, it is mapped to +
d, where 2d is the distance betw
een the symbols. W
e will need 2 rvs, one of the
real part and one for the imaginary part. C
hose “d” to ensure Eb=
1. Generate 10
5sym
bols to measure the S
ER
over the same range of SN
Rs as in part (c). P
lot your results back in Fig-2. C
omm
ent.
(e) Repeat part (c) above for the 16-Q
AM
signal set. Call this Fig-3.
(f) Repeat part (d) for the 16-Q
AM
signal set. Chose the rv to sym
bol mapping appropriately and explain how
you did it. Plot this result back in Fig-3.
(g) With G
ray-coding done, compare the sim
ulated bit error rate (BE
R) of Q
PS
K and 16-Q
AM
. (Hint: w
hile you are measuring the S
ER
in parts (e) and (f), you can also com
pute the BE
R for both the m
odulations). Plot these tw
o results together in Fig-4, again for Eb=
1 in both cases. Note that for com
puting BE
R, you w
ill
Instructions:S
ubmit only a single P
DF file for the report. N
ame the file as “rollnum
ber-assignment1-report.pdf”.
Your w
orking code, properly comm
ented, mu
st be inclu
ded
as Annexure-1 to the P
DF file, and also separately
sent by email to the T
As. Y
our working code can be nam
ed “rollnumber-assignm
ent1-code.m”. P
lease see other instructions, if any, in the W
hatsApp group.