Transmitted signal reaches Rx through several paths, each path has: a) attenuation. b) phase. c) delay.
Channel is described by impulse response
Parameters of the mobile multipath wireless channel:
EXA: Sunday 2/10/13 11:03 AM
Freq. selective fading: | H(w) | is variable with (w) over the BW of the signal.
Power delay profile: is the spatial average
RMS delay spread of the channel:
EXA: find rms delay spread for the channel with power delay profile.
Solution: Flat fading : | H(w) | is approximately constant over the BW of signal. Coherence BW ( Bc ): Is the range of freqs. over which the ch. H(w) is consedered flat, in other words ( i.e. ), two freqs. are highly correlated if f2 - f1 < Bc
In the previous EXA: Tuesday 2/12/13 10:58 AM
Follow previous EXA:- b) would this channel be suitable for AMPS (BW = 30 kHz) GSM (BW = 200 kHz)
Solution
Can operate properly without equalizer (flat fading)
GSM: BWsig > Bc needs equalizer (Freq. selective)
Freq. spreading (Doppler effect)
Max. Doppler shift
Coherence time (Tc): Is the time at which h(t) is approximately unchanged.
EXA: a mobile Rx at fc = 2.2 GHz moving at speed v = 20 km/hr Find Tc.
Solution
Thursday 3/5/13 11:03 AM
Conventional methods for ch. fading: 1) Time - selective fading:- Solution: ch. coding & interleaving.
Adaptive modulation & coding: Is one of the key - technology in 3G/4G system. Enables spectrally efficient transmission over time varying channels.
Basic Concepts: - To estimate the ch. gain ( ) at Rx and feedback to the Tx. - Tx dynamically changes:- 1) Modulation, 2) Power & 3) Coding according to ch. condition ( ) within each frame (frame length < Tc). - The goal is to max. spectral efficiency.
Block diagram: Adaptive modulation & coding Power control Demodulation Channel estimation Adaptive System Generated by CamScanner Sunday 2/17/13 11:02 AM
Bs = 1/Ts (Nyquist pulse.) P : Average Tx power.
Define
SNR (Signal-to-Noise power ratio) at Rx if P power is transmitted.
Adaptation params: The Tx power for a specific ( ) The data rate for a specific ( )
Finding Rayleigh channel
Example on rate-variation with fixed power: Consider adaptive modulation using:- 1) QPSK (M = 4) 2) 8-PSK (M =8) To meet a target prob. of bit error Pb = 10^-3
Tx works as follows: 1) if received SNR is sufficient for 8-PSK to have Pb = 10^-3 , use 8-PSK. 2) if received SNR is sufficient for QPSK to have Pb = 10^-3 , use QPSK (but not sufficient for 8-PSK to have Pb = 10^-3) 3) if received SNR is not sufficient for QPSK / 8-PSK to meet target Pb DO NOT SEND DATA ... EXA: Find:- a) find SNR regions ( ) assigned to each modulation. (QPSK, 8-PSK, No data) b) find spectral eff. R/B for this method. b) compare to non-adaptive.
Solution Tuesday 2/19/13 11:04 AM
Follow EXA:-
1) Variable power technique: we vary Tx-power so that SNR at Rx is constant.
Adaptation:-
(Non adaptive >> no feedback >> high error.) Adaptation method:-
There is synchronization between Tx & Rx, so Don't send: makes sense. Thursday 2/21/13 11:05 AM
Truncated channel inversion: For Rayleigh channel:-
Select threshold
EXA: find the power adaptation for BPSK modulation for a target BER Pb = 10^-3. Assuming Rayleigh channel with b) Find outage prob. (Pout) c) Find spectral eff. R/B. Solution
2) Variable - rate, variable power schemes: Rate/Power is selected to maximize spectral eff. (R/B), while meeting target Pb. For good ch. But, optimized?
Derivation:-
The goal is to maximize R/B:- Sunday 2/24/13 11:24 AM
(Water Filling) Tuesday 2/26/13 11:02 AM
EXA: assume a fading ch. with an average power p, and if p was sent by Tx then Rx SNR has PDF
Assume target BER =
Solution
Spectral eff. of non-adaptive systems: BER is obtained by averaging prob. of error (AWGN) over fade distribution Which M satisfies (Pb)
Rayleigh ch. Thursday 2/28/13 11:05 AM
EXA (previous) Find the spectral eff. for non-adaptive system.
Solution
3) channel inversion with fixed rate: Assume MQAM, we achieve fixed SNR that satisfies Pb. Choice of M?
We need to max. R/B
Power assignment
Rate eff. (fixed)
EXA: assume the channel
Find power/rate policy and achievable spectral eff., if we use truncated ch. inversion with fixed rate.
Solution
Sunday 3/3/13 11:04 AM
Discrete - rate adaptation: MQAM , power is cont., M is discrete, Mo = 0 (no transmission) ,M1 = 2 ,M2=4,
Chapter Orthogonal Frequency Division Multiplexing (OFDM) OFDM is now widely used in standards for broadband comm. Due to its ability to work efficiently in freq. - selective chs.
Basic idea: Is to divide a high rate data stream into N - parallel substreams, each modulated on a separate carrier, which are all orthogonal
OFDM increases symbol duration (Ts) by a factor of (N) so that:- Thursday 3/7/13 11:02 AM
Multicarrier comm. system:
Disadvantage:- High complexity in implementation of LPFs, multipliers ... Especially for large (N).
OFDM using complex multipliers:
Standard OFDM (using IFFT - FFT): Instead of sending continuous time x(t) (band-limited to BW) we can send sampled signal (Nyquist rate).
Sample period
N-samples will be sufficient:- 1- reconstruct x(t) at Rx. 2- more importantly to recover the data Sunday 3/10/13 10:57 AM
(Standard OFDM):-
OFDM with cyclic prefix: Cyclic prefix: extending the OFDM symbol by copying it's last part into it's beginning.
Reasons for CP:- 1- prevent interference of adjacent OFDM symbols. 2- converts Linear-convolution into circular-convolution to maintain orthogonality between channels. Tuesday 3/12/13 11:03 AM
Detection of the OFDM signal:
Assume timing error of ( ): Thursday 3/14/13 11:04 AM
Error in starting point of OFDM symbol:
Case 1:- Extimated starting point is the exact staring point. orthogonality is preserved. No ICI.
Case 2:- Estimated start is before exact start but after length L. orthogonality is preserved. No ICI.
Case 3:- Estimated start is within 0 to L.
Case 4:- Estimated start is after exact start.
- case 3 & 4 :- orthogonality is destroyed. ICI + ISI.
Block diagram:
EXA: given OFDM system with:-
Find: a) sub channel BW. b) overhead ratio of CP. c) Data rate if Generated by CamScanner
Solution Sunday 3/17/13 11:03 AM
EXA: IEEE 802.11a WLAN
Spectral eff. of OFDM: Ts: OFDM symbol duration. N: no. of subcarriers. Tch: sampling period = Ts/N. PSD of OFDM and reduction of out-of-band radiation:
Reduction techniques: 1) Digital filtering Disadvantages:- 1- complexity. 2- filters impulse response is added to overall ch.
2) Raised cosine windowing.
3) Zero carriers at the edges.
Tuesday 3/19/13 11:05 AM
Design guidelines for OFDM system: (choice of OFDM parameters) 1) Gaurd interval Tcp (Duration of CP)
2) Ts
Effects of wireless ch. on OFDM system: Time varying >> orthogonality is destroyed. 1) Carrier freq. offset (CFO). 2) Doppler spread. 3) Sampling clock offset.
Time varying ch. effect:
OFDM Tx - signal
The demodulated signal: Fixed channel:
Effect of CFO: Generated by CamScanner Tuesday 3/21/13 11:06 AM
Case 1:
Case II: We get circular shift of detected data. Generated by CamScanner Sunday 3/24/13 11:00 AM Generated by CamScanner Generated by CamScanner Generated by CamScanner Tuesday 3/26/13 11:05 AM
Performance optimization of OFDM systems: Loading of parallel channels:- 1) Rate maximization loading. 2) Energy minimization loading.
1) Rate maximization loading: Given fixed total power P Maximize bit-rate. Pi: power assigned to ith ch. P : total power
Np: noise power in each ch. Assume MQAM with target prob. bit error (Pb) Rx - SNR =
Maximize No. Of bits/OFDM symbol: The Rx-SNR at the ith ch. if unity power is transmitted. Subject to the constraint Rate maximization algorithm:- Margin maximization: Minimize P subject to fixed target bit rate (fixed BER Pb.) Algorithm: Thursday 3/28/13 11:06 AM
EXA: 4- ch. (N=4) OFDM system with a two-tap ch.
Noise power in each ch. = 0.4 The available Tx power P = 8 Target BER Pb =
Det. The optimum (waterfilling) power assignment for each subcarrier and the total no. of bits/OFDM symbol. Power Assaignment:-
Rate Assaignment:-
Channel Estimation: Generated by CamScanner Generated by CamScanner Sunday 3/31/13 11:06 AM
Estimation of ch. requires sending pilot symbols (known data) at some subcarriers. Interpolation is used to estimate the ch. at other subcarriers.
Pilot structure: 1) block type:- Pilots transmitted on all subcarriers for one OFDM symbol.
St must satisfy
Time - domain interpolation is used to estimate the ch. at other OFDM symbols. Suitable for highly freq-selective fading, but not fast-fading.
2) comb. type:-
Freq. domain interpolation is used to find ch. estimation at other subcarriers. Sf must satisfy
Suitable for fast - fading, but not for high - selectivity in freq.
3) Lattice type:- Tuesday 4/2/13 11:06 AM
Basic definitions: Training data (pilot):-
Least - squares ch. estimation: We want to find
that minimizes
Minimum Mean Squared Error (MMSE) estimation: Used to find:- a) chs. at the same pilots. b) chs. at other chs. Given LS - estimation of pilot chs. : Define
Define the MMSE estimation:- Ch. vector to be interpolated. To estimate some chs. & improve other chs. Sunday 4/7/13 11:04 AM
MMSE ch. estimator (Weiner estimator):
we need to minimize
Minimizing is eqnt. to solving:- Elements of Power Delay Profile (Principle of orthogonality) Generated by CamScanner Tuesday 4/9/13 11:05 AM This example is an application for video broadcasting ======>> Generated by CamScanner Generated by CamScanner Tuesday 4/16/13 11:06 AM
Decision directed channel estimation: Used with block type plots. For slowly varying ch. 1) ch. at the kth subcarrier is estimated by LS.
2) assuming ch. has not changed (approx. the same) in the next symbol, we can estimate data at t = (n+1) using the estimate at (n)
3)
4) process is repeated for OFDM symbols until the next pilot comes.
Linear / 2nd order interpolation for comp type pilots:
2nd order interpolation:
DFT - band ch. estimation: This method improves LS-estimate. Synchronization in OFDM systems:
1- Sample time offset (STO):- STO width means finding the starting point of the OFDM symbol.
2- carrier freq. offset (CFO) estimate:- Error in the local oscillator compensated to Tx oscillator.
Once the STO, CFO are estimated they will be compensated for & Rx is now synchronized. Thursday 4/18/13 11:21 AM
1) Estimation of STO: 1- Using cyclic prefix:- We use the similarity of CP with the end portion of the OFDM symbol. We find point of maximum similarity. Let estimated STO be Absolute deference here is within CP Tuesday 4/23/13 11:16 AM
Solution of 1st exam: Q1:- Q2:- Generated by CamScanner Thursday 4/25/13 11:01 AM Generated by CamScanner Generated by CamScanner Generated by CamScanner Tuesday 4/28/13 11:04 AM
Channel capacity (waterfilling):
Algorithm:
EXA: find the ch. capacity for:-
Solution
EXA: find the capacity for:-
Solution
Chapter Space - Time Coding Transmitt diversity:-
Time delay diversity:-
Block (Alamouti) ST (Space Time) code: Tuesday 4/30/13 11:04 AM
Real space - time coding: Code rate = 1 Square matrix (Quantization to nearest pt. in constellation)
R = K
No. Of symbols transmitted. No. Of time instants. Non-Square: Generated by CamScanner