- 1 - CDMA Mobile Radio Networks Elvino S. Sousa Department of Electrical and Computer Engineering University of Toronto Canada ECE1543S - Spring 1999
- 2 - CONTENTS Basic principle of direct sequence (DS) spread spectrum Spreading code sequences Types of DS modulators Synchronization Symbol error probability Multiple access capability Forward error correction Frequency hopping Reed Solomon hopping sequences CDMA cellular networks Call Blocking CDMA capacity equation Signal propagation Rake receiver Power Control Performance in multi-path fading Interleaving IS-95 standard European CODIT project Summary
- 3 - SPREAD SPECTRUM Definition: A signaling scheme which has a bandwidth that is much larger than the minimum bandwidth typically required for the given data rate (typical bandwidth = data rate). Properties: interference rejection anti-jamming communications anti-multi-path fading (frequency diversity) low probability of intercept secrecy? multiple access capability frequency re-use (cellular architecture) Types of Spread Spectrum. There are many different types of spread spectrum techniques. Mathematically they are all based on the same principle. From implementation point of view the main types are: direct sequence (DS) frequency hopping (FH, slow, fast) time hopping hybrid FH/DS
- 4 - DIRECT SEQUENCE Data Source Data Baseband SS Signal Channel ct () Acos( ω c t) PN code (Spreading Code) It () Interference ct () Acos( ω c t) t = kt kt ( k 1)T +1-1 +1-1 +1-1 1 0 0 1 data signal spreading code baseband spread spectrum signal
- 5 - Power Spectral Density Data signal Baseband Spread spectrum signal Processing Gain: N = T ----- T c 1 T --- Frequency 1 ---- T c Power spectral densities in the channel Narrow Band Interference Bandpass SS Signal ω ----- c 2π Frequency
- 6 - Despreading Input to the integrator (or low pass filter) Low pass filter (Integrator) Collapsed data signal Rejected interference Detector noise Spread Interference 0 -- 1 T Frequency ----- 1 T c
- 7 - Spreading Code Sequences short sequences - typical sequence period equal to data symbol period. long sequences - sequence period much longer than data symbol period - model the sequence as random. Sequence Generation: - use a shift register with feedback connections. Example: PN sequence of period 63. Galois Field Form Recurrence Relation Form Generator Polynomial: P x ( ) = x 6 + x + 1
- 8 - Properties of PN Sequences Feedback connections may be linear or non-linear. A shift register with n stages has 2 n possible states, therefore maximum sequence period = 2 n. For linear sequences 0 state is not allowed, therefore maximum sequence period = 2 n 1. Linear PN sequences sequence period is a divisor of 2 n 1. maximum period = 2 n 1. sequences with maximum period exist for all n (called m-sequences). m-sequences have good randomness properties (Golomb postulates): R1 - in one period: # 1 s = # 0 s + 1 = 2 n 1. R2 - the number of runs of length r equals 2 n r (a run is a sequence of symbols of constant value, delimited by two symbols of the opposite value). R3 - convert 0 -> 1, 1 -> -1, then the periodic autocorrelation of the PN sequence ( ) is given as follows: R( λ) where p = 2 n 1. p 1 = -- b p i b i + λ = i = 1 b i 1 if λ = 0 1 -- if 0 < λ < p p R( λ) 1 1 -- p λ
- 9 - Feed-Back Connection Polynomials For maximum period the polynomial must be a primitive polynomial in GF( 2 n ). In general if the polynomial is irreducible over GF( 2) the period equals the order of the roots in GF( 2 n ). Number of primitive polynomials over GF( 2) : φ( 2 n 1) ---------------------- n where φ( x) is the Euler φ function = number of integers less that x that are relatively prime to x. n number of primitive polynomials period representative polynomial 2 1 3 3 2 7 4 2 15 5 5 31 6 3 63 7 18 127 8 16 255 x 2 + x + 1 x 3 + x + 1 x 4 + x + 1 x 5 + x 2 + 1 x 6 + x + 1 x 7 + x 3 + 1 x 8 + x 4 + x 3 + x 2 + 1
- 10 - Types of Modulators BPSK Transmitter: dt () chip shaping filter ct () cos( ω c t) Receiver: dt () T ( k 1)T cos( ω c t) ct ()
- 11 - QPSK1 Transmitter: dt () S/P c i () t c q () t cos( ω c t) sin( ω c t) Receiver: cos( ω c t) sin( ω c t) kt ( k 1)T c i () t c q () t P/S kt ( k 1)T
- 12 - OQPSK1 Transmitter: dt () S/P c i () t c q () t cos( ω c t) sin( ω c t) T ----- c 2 Receiver: kt ( k 1)T cos( ω c t) c i () t sin( ω c t) T c /2 c q () t P/S kt ( k 1)T
- 13 - QPSK2 Transmitter: dt () c i () t c q () t cos( ω c t) sin( ω c t) Receiver: cos( ω c t) sin( ω c t) c i () t c q () t kt ( k 1)T OQPSK2: Insert a half-chip delay in the quadrature branch of QPSK2 scheme.
- 14 - Synchronization The receiver requires a replica of the PN code, with the correct clock phase, at the receiver in order to despread the signal. Typically this code must be derived from the received information signal. The process of synchronizing to the transmitter PN code consists of two steps: Acquisition (coarse synchronization). Tracking (fine synchronization). A receiver typically consists of acquisition circuits, tracking circuits, and demodulator circuits. Acquisition Tracking Demodulation Operation: acquisition, tracking + demodulation, loose tracking, acquisition, tracking + demodulation,...
- 15 - Acquisition Serial maximum-likelihood search technique: Correlate the received signal with the local PN code with all possible code clock phases. Choose the clock phase which yields the maximum correlation. Advantage: optimum acquisition technique Disadvantage: Slow if the code period and uncertainty in the code phase are large. Received signal plus noise BPF Square-Law Envelope Detector t t τ d Sample at t = jτ d Z j Local PN Code Generator Update code phase by τ T c at t = jτ ; d j 1 2 q = q = # positions examined T c = chip period,,, Estimated code phase Store Z j ; choose local PN code phase corresponding to max Z j j
- 16 - Code Search Strategies To shorten the time to acquisition an incomplete search is performed according to some strategy: e.g. single-dwell, multiple-dwell, linear, zig-zag, expanding window, discrete, continuous. Single dwell time PN acquisition system with non-coherent detection Correlate with the local PN code at various clock phases until the correlation exceeds a pre-specified threshold. Declare synch. when the threshold is reached. Advantage: Shortens the time to acquisition Disadvantage: Not optimum; may stop searching too soon. Received signal plus noise BPF Square-Law Envelope Detector t t τ d Sample at t = jτ d Z j Local PN Code Generator PN code phase update No Yes Threshold Comparison
- 17 - Equivalent Low-Pass Representation of the Single Dwell Time PN Acquisition System Received signal plus noise cos( ω c t + θ) LPF LPF (). 2 (). 2 t t τ d Sample at t = jτ d sin( ω c t + θ) Local PN Code Generator PN code phase update No Yes Threshold Comparison
- 18 - Rapid Acquisition by Sequential Estimation (RASE) Advantage: Very fast acquisition under high SNR operation. Disadvantage: Speed advantage is lost as the SNR decreases. In-lock detector PN code feedback logic Track Received signal plus noise PN code chip detector Load 1 2 3 n Examination period generator Load/Track Logic
- 19 - Tracking One Method: Non-coherent delay-locked loop PN reference signal ct ( τ) ε + () t y + () t BPF. () 2 y 2 + () t signal plus noise Delay δ c( t τˆ + δ) PN Code Generator c( t τˆ δ) VCO Loop Filter ε() t _ + ε - () t BPF. y - () t () 2 y2 - () t Branch autocorrelations: R + ( ε) R - ( ε) ε δ T c + δ ε Discriminator Characteristics - S Curve D( ε) = R - ( ε) R + ( ε) D( ε) ε