Institute for Experimental Mathematics Ellernstrasse Essen - Germany DATA COMMUNICATION 2-dimensional transmission A.J. Han Vinck May 1, 2003
University Duisburg-Essen digital communications group 2 Content we describe orthogonal signaling 2-dimensional transmission model
University Duisburg-Essen digital communications group 3 „orthogonal“ binary signaling 2 signals S 1 (t) S 2 (t) in time T Example: Property:orthogonal energy E TT
University Duisburg-Essen digital communications group 4 Quadrature Amplitude Modulation: QAM S(t)
University Duisburg-Essen digital communications group 5 QAM receiver +/- 1/0 +/- 1/0 r(t) r(t) = S(t) + n(t) Note: sin(x)sin(x) = ½ (1 – cos (2x) ) sin(x)cos(x) = ½ sin (2x)
University Duisburg-Essen digital communications group 6 about the noise
University Duisburg-Essen digital communications group 7 about the noise Conclusion: n 1 and n 2 are Gaussian Random Variables zero mean uncorrelated (and thus statistically independent (f(x,y) =f(x)f(y) ) with variance 2.
University Duisburg-Essen digital communications group 8 Geometric presentation (1)
University Duisburg-Essen digital communications group 9 Geometric presentation (2) ML receiver: find maximum p(r|s) min p(n) decision regions
University Duisburg-Essen digital communications group 10 performance From Chapter 1: P(error) =
University Duisburg-Essen digital communications group 11 extension 4-QAM 2 bits 16-QAM 4 bits/s Channel 1 Channel 2
University Duisburg-Essen digital communications group 12 Geometric presentation (2) 11 22 transmitted received noise vector n The noise vector n has length |n| = ( 2 2 ) ½ n has a spherically symmetric distribution! equal density
University Duisburg-Essen digital communications group 13 Geometric presentation (1) r‘ r d/2 Prob (error) = Prob(length noise vector > d/2)
University Duisburg-Essen digital communications group 14 Error probability for coded transmission The error probabiltiy is similar to the 1-dimensional situation: We have to determine the minimum d 2 Euclidean between any two codewords Example: C C‘ d 2 Euclidean =
University Duisburg-Essen digital communications group 15 Error probability The two-code word error probability is then given by:
University Duisburg-Essen digital communications group 16 modulation schemes On-offFSK 8-PSK 3 bits/s 16-QAM 4 bits/s 4-QAM 2 bits 1 bit/symbol
University Duisburg-Essen digital communications group 17 transmitted symbol energy energy: per information bit must be the same FSK
University Duisburg-Essen digital communications group 18 performance From Chapter 1: P(error) = d/2 FSK
University Duisburg-Essen digital communications group 19 Coding with same symbol speed In k symbol transmissions, we transmit k information bits. We use a rate ½ code In k symbol transmissions, we transmit k bits ML receiver:
University Duisburg-Essen digital communications group 20 Famous Ungerböck coding In k symbol transmissions transmit We can transmit 2k information bits and k redundant digits In k symbol transmissions transmit 2k digits Hence, we can use a code with rate 2/3 with the same energy per info bit!
University Duisburg-Essen digital communications group 21 modulator info encoder cici c i {000,001,010,...111} Signal mapper 2323
University Duisburg-Essen digital communications group 22 example transmit or Parity even Parity odd Decoder: 1) first detect whether the parity is odd or even 2) do ML decoding given the parity from 1) Homework: estimate the coding gain
University Duisburg-Essen digital communications group 23 Example: Frequency Shift Keying-FSK Transmit:s(1):= s(0):= Note: FSK
University Duisburg-Essen digital communications group 24 Modulator/demodulator S(t) m m r(t) m Select largest demodulator modulator
University Duisburg-Essen digital communications group 25 Ex: Binary Phase Shift Keying-BPSK Transmit:s(1):= s(0):= m > or < 0? m‘
University Duisburg-Essen digital communications group 26 On-off BFSK BPSK Modulation formats
University Duisburg-Essen digital communications group On-off BPSK QPSK Eb/N0 dB Error rate PERFORMANCE