ECE 4710: Lecture #10 1 Digital Signaling What is appropriate way to mathematically represent the waveform of a digital signal? What is the bandwidth of the digital signal? BW depends on pulse shape used to represent digital data Only indirectly related to bandwidth of analog signal bandwidth via sampling frequency f s Digital waveform can be represented by series summation of N orthogonal functions N is dimension of orthogonal function set = # of ( t ) functions required to represent all possible waveforms for digital signal w k represents the digital data (e.g. 101 w 1 = 1, w 2 = 0, w 3 = 1)
ECE 4710: Lecture #10 2 Orthogonal Functions What is orthogonal? satisfies mathematical condition Example: sin( t ) and cos( t )
ECE 4710: Lecture #10 3 Orthogonal Functions Orthogonal Another word is “perpindicular” »Sine and cosine are 90° out of phase In complex domain »Orthogonal characteristic of sine/cosine »Cosine Real axis »Sine Imaginary axis »Used to represent vector R Uniqueness of orthogonal characteristic enables the vector representation Many other types of orthogonal function sets |R|cos |R|sin Real Im
ECE 4710: Lecture #10 4 Symbol & Bit Rate For N dimension waveform set transmitted over T 0 seconds: Symbol Rate = D = N / T 0 (symbols/sec or sps) »Also called baud rate outdated »Please use symbol rate (sps) in this class Bit Rate or Data Rate = R = n / T 0 (bits/s or bps) If w k ’s have binary values then n = N and D = R »2 states only per symbol binary signaling If w k ’s have more than 2 possible states and D R
ECE 4710: Lecture #10 5 Vector Representation Orthogonal function space can be represented in vector space by where w is an N dimensional vector and the set { j } is orthogonal set of N directional vectors Shorthand notation for w is row vector Note book uses bold w for vector representation
ECE 4710: Lecture #10 6 Vector Representation Three-bit binary signal s(t) represented by 3-bit waveform Let p(t) So t 5 V T T 0 = 3T t 5 V T Bit-Shape Waveform Functional Space
ECE 4710: Lecture #10 7 Vector Representation Orthogonal Function Set t 5 V T p(t)p(t) t T 2T 3T p1(t)p1(t) t 5 V T 2T 3T p2(t)p2(t) t 5 V T 2T 3T p3(t)p3(t)
ECE 4710: Lecture #10 8 Vector Representation Orthogonal Vector Space N = 3 dimensions 2 N = 8 possible messages for each symbol
ECE 4710: Lecture #10 9 Digital Signal BW Lower bound for digital signal BW Lower bound only achieved for sin( x )/ x pulse shape Other real pulse shapes will have larger BW Binary Signal Example M = 256 possible message & n = 8-bit binary words T = 1 msec so T 0 = 8 msec Example message = so
ECE 4710: Lecture #10 10 Binary Signal Example Bit Rate = R = n / T 0 = 8 / 8 ms = 1 kbps Symbol Rate = D = R = 1 ksps since it is a binary signal Rectangular Pulse Shape, T b = 1 ms sin(x)/x Pulse Shape, T b = 1 ms
ECE 4710: Lecture #10 11 Binary Signal BW Rectangular Pulse FNBW = 1 / T = 1 / 1 msec = 1 kHz Digital source info transmitted with digital waveform Sin( x ) / x Pulse Smooth rounded corners have much less frequency content Digital source information transmitted with analog waveform Pulse shape has no ISI if sampled exactly at midpoint of bit period see sampling points in previous figure Absolute BW = minimum BW = 0.5 D = 500 Hz Other Pulse Shapes Filter rectangular pulses to reduce BW Studied next
ECE 4710: Lecture #10 12 Multi-Level Signaling Multi-level signaling Binary signals have L = 2 states/symbol »“0” = 0 V and “1” = +5V Multi-level signaling has L > 2 states/symbol »# bits / symbol = log 2 ( L ) Two possible benefits: 1)For same symbol period, if L then # bits per unit time data rate is increased OR 2) If L we can increase the symbol period to maintain the same data rate BW 1 / T s so BW will be reduced
ECE 4710: Lecture #10 13 Multi-Level Signaling Binary to Multi-Level Conversion for L = 4 Binary InputOutput Level Example message = -3, -1, +3, +1 Same message as previous binary signal example
ECE 4710: Lecture #10 14 Multi-Level Signaling Bit Rate = R = n / T 0 = 8 / 8 ms = 1 kbps Symbol Rate = D = N / T 0 = 4 / 8 ms = 500 sps N = 4 Dimensions { w 1, w 2, w 3, w 4 } = {-3, -1, +3, +1} Multi-Level Rectangular Pulse Shape Multi-Level sin(x)/x Pulse Shape T s = 2 msec
ECE 4710: Lecture #10 15 Multi-Level Signal BW Rectangular Pulse FNBW = 1 / T s = 1 / 2 msec = 500 Hz Sin( x ) / x Pulse Absolute BW = minimum BW = 0.5 D = 250 Hz BW’s are 2 smaller than same message for binary signal Data rate kept the same Symbol period increased by factor of 2 BW Alternate approach would be to keep same BW and increase data rate by factor of 2