I want to draw the complex-valued baseband signal with real $S_{BB,I}(t)$ and imaginary part $S_{BB,Q}(t)$ for the following bit sequence: 011111001000

I have looked around and found on wikipedia such a timing diagram but for QPSK: Timing diagram for QPSK

I understand that the bit sequence of a QPSK is grouped in pairs of two, the first bit of the pair is assigned to I, the second part of the pair is assigned to Q. If I=1, the sine starts at the top. If Q=1, the cosine rises.

How would I need to draw this diagram for an 8-PSK? The bits are grouped in pairs of three. So my bit sequence is 011,111,001,000. I am puzzled how to assign the pairs of three to only two signals I and Q.

  • $\begingroup$ The wikipedia image out of context is confusing; this is not the value of the symbols in baseband. As the text above the image on the wikipedia page says: "The modulated signal is shown below for a short segment of a random binary data-stream. The two carrier waves are a cosine wave and a sine wave"; carrier waves, not baseband. If you're not sure about the difference, please read up on complex baseband first. $\endgroup$ Dec 15, 2016 at 17:58

2 Answers 2


You're making your life actively harder by considering I and Q separate instead of considering them to be the real and imaginary part of a complex signal $S = I + jQ$.

If you then do that, you'd simply get a table with 8 rows, one for every constellation point. Notice that for 8-PSK, you simply divide the full unit circle ($2\pi$) into 8 equal angles – hence, the complex value of all these points is simply $e^{j\angle(\text{constellation point})}= e^{j\frac{2\pi}{8}n}$; I arbitrarily chose that 000 should be mapped to $n=0$, 001 to $n=1$ and so on. There's no reason that this is the right mapping – all 8-PSK says is "there's 8 constellation points on equally distant points on the unit circle of the complex plane".

  • 000$_b\equiv 0$:  $e^{j\frac{2\pi}8 \cdot 0}=1+0j$
  • 001$_b\equiv 1$:  $e^{j\frac{2\pi}8 \cdot 1}=\frac{\sqrt2}2+\frac{\sqrt2}2j$
  • 010$_b\equiv 2$:  $e^{j\frac{2\pi}8 \cdot 2}=0+1j$
  • 011$_b\equiv 3$:  $e^{j\frac{2\pi}8 \cdot 3}=-\frac{\sqrt2}2+\frac{\sqrt2}2j$
  • 100$_b\equiv 4$:  $e^{j\frac{2\pi}8 \cdot 4}=-1+0j$
  • 101$_b\equiv 5$:  $e^{j\frac{2\pi}8 \cdot 5}=-\frac{\sqrt2}2-\frac{\sqrt2}2j$
  • 110$_b\equiv 6$:  $e^{j\frac{2\pi}8 \cdot 6}=0-1j$
  • 111$_b\equiv 7$:  $e^{j\frac{2\pi}8 \cdot 7}=\frac{\sqrt2}2-\frac{\sqrt2}2j$

You'd then just look up the real and imaginary part from that table. The simple "bit maps to I (or Q)" simply doesn't work for anything but BPSK and QPSK.

  • $\begingroup$ thank you, where does your sine and cosine curve for the value $011 \equiv -0.7071 + 0.7071j$ start? I assume that sine starts at $-0.7071$ at the beginning of the interval T and cosine starts at $0.7071$ at the the beginning of the interval T. $\endgroup$
    – autoship
    Dec 15, 2016 at 13:48
  • $\begingroup$ There's no cosine curve in a constellation mapping – that happens due to signal shaping through limited channel bandwidth or pulse shaping $\endgroup$ Dec 15, 2016 at 14:26
  • $\begingroup$ you're confusing 8-PSK (a constellation) with a 8-PSK-modulated carrier $\endgroup$ Dec 15, 2016 at 14:26
  • $\begingroup$ In baseband, there's no cosines/sines here at the symbol periods – it's just the values. $\endgroup$ Dec 15, 2016 at 14:27
  • $\begingroup$ my exercise asks me to draw on two diagrams: 1st diagram: vertical axis: $i(t)$, horizontal axis: $t[\mu s]$ 2nd diagram: vertical axis: $q(t)$, horizontal axis: $t[\mu s]$ It looks like I have to draw something like this mathworks.com/help/comm/ug/qpsk_transition_timing.png $\endgroup$
    – autoship
    Dec 15, 2016 at 15:07

You have to map the grouped binary values to the constellation of 8psk from the constellation you will get the value of I and Q.

000 - -1/sqrt(2) - j(1/sqrt(2)) 001 - -1 010 - j 011 - -1/sqrt(2) + j(1/sqrt(2)) 100 - -j 101 - 1/sqrt(2) - j(1/sqrt(2)) 110 - 1/sqrt(2) + j(1/sqrt(2)) 111 - 1


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.