EC22603 - Wireless Communication

Assignment
Haresh Krishna G S
ECE - ’A’
2127220701043

May 22, 2025

FT/GN/61A/00/02.07.18
B.E. ECE VI Semester
Marks: 40 (10 per question)
Course Outcomes: CO4, CO5
RBT Level: Analyze

Question 1: Adaptive Algorithms for DSP Chip

If a digital signal processing chip can perform one million multiplica- tions per second,
analyze the best algorithms which provide rate of convergence as a dependable factor
of data rate and coherence time. Marks: 10 CO: 4 RBT Level: Analyze

1.1 Introduction
Adaptive filtering algorithms are essential in digital wireless communication systems
to mitigate time-varying channel effects such as multipath fading. The coherence time
(Tc) defines the duration over which the channel remains stable, while the data rate (fs)
determines the number of symbols processed per second. A DSP chip with a capacity
of one million multiplications per second (1M MPS) imposes computational
constraints. This analysis evaluates three algorithms— Least Mean Squares (LMS),
Normalized LMS (NLMS), and Recursive Least Squares (RLS)—based on their
convergence rates and suitability for varying Tc and fs.

1.2 Key Parameters

• Coherence Time (Tc): Duration where the channel is constant, approxi- mated
as Tc ≈ f1d , where fd = v λis the Doppler frequency, v is velocity, and λ is
wavelength. For example, at 900 MHz (λ ≈ 0.33 m) and v = 60 km/h =
16.67 m/s, fd ≈ 50 Hz, so Tc ≈ 20 ms.

1
 • Data Rate (fs): Symbol rate (e.g., ksps), affecting adaptation steps.

• Number of Symbols (N ): N = Tc · fs, symbols available for adaptation.

• Convergence Rate: Speed of error minimization, critical for fast-fading
channels.
• DSP Constraint: 1M MPS limits filter length (L) and fs.

1.3 Least Mean Squares (LMS) Algorithm

Overview
LMS uses stochastic gradient descent to minimize mean square error (MSE), known for
simplicity and low complexity.

Mathematical Foundation
Weight update rule: w(n + 1) = w(n) + µe(n)x(n)

where w(n) is the weight vector, µ is step size, e(n) = d(n) − wT (n)x(n) is the error, d(n) is
the desired signal, and x(n) is the input vector of length L.

Convergence Rate

Time constant: τ ≈ 1 , where σmin is the smallest eigenvalue of the input cor-
relation matrix. Convergence requires 100-200 iterations, suitable for large N (slow-
fading channels).

Computational Complexity
Per symbol: 2L+1 multiplications (L for output, L for weight update, 1 for error). For fs =
50 ksps, L = 9, total = 50, 000 · 19 = 950, 000 MPS, within limit.

Use Case
Best for stable channels with low to moderate fs.

1.4 Normalized LMS (NLMS) Algorithm

Overview
NLMS normalizes the step size based on input signal power, improving stability and
convergence.

2
Mathematical Foundation
Weight update:
µ
w(n + 1) = w(n) + e(n)x(n)

where ϵ prevents division by zero.

Convergence Rate

Converges in 50-100 iterations, suitable for moderate Tc.

Computational Complexity

Per symbol: 3L + O(1) multiplications. For fs = 50 ksps, L = 6, total = 50, 000 · 19 ≈ 950,
000 MPS, feasible.

Use Case

Balances performance and complexity for moderately varying channels.

1.5 Recursive Least Squares (RLS) Algorithm

Overview

RLS minimizes weighted least squares error, offering fast convergence at high
computational cost.

Mathematical Foundation
Key equations:
P(n − 1)x(n) k(n) = λ +
x (n)P(n − 1)x(n)
T

w(n) = w(n − 1) + k(n)e(n)

1
( T (n)P(n − 1))
P(n) = P(n − 1) − k(n)x
λ
where λ is the forgetting factor (0.95-1).
Convergence Rate

Converges in ∼ 2L iterations (e.g., 10 for L = 5), ideal for small Tc.

Computational Complexity

Per symbol: 2L2 +O(L) multiplications. For fs = 20 ksps, L = 4, total = 20, 000 · 32 = 640,
000 MPS, feasible.
Use Case
Best for high fs and fast-fading channels with small L.

1.6 Algorithm Selection

Scenario Recommended Algorithm

Rapidly changing channel, high fs RLS (if L small)

Moderate channel variability NLMS Stable channel,
low fs LMS

Table 1: Algorithm selection based on system conditions

1.7 Example Scenario

For fs = 50 ksps, Tc = 10 ms:

• N = 0.01 · 50, 000 = 500 symbols.

• LMS: With L = 8, µ = 0.01, convergence in ∼ 100 iterations, MPS = 50, 000 · 17 =

850, 000.

• NLMS: Convergence in ∼ 50 iterations, L = 6, MPS = 950, 000.

• RLS: Convergence in ∼ 10 iterations, L = 4, MPS = 640, 000.

Diagram Suggestion: LMS convergence curve (error vs. iterations).

1.8 Conclusion
LMS suits slow-fading channels and low fs due to low complexity. NLMS balances
performance for moderate Tc, while RLS excels in fast-fading channels but is
constrained by the 1M MPS limit unless fs and L are low. Selection depends on N = Tc ·
fs and computational feasibility.

4
Question 2: Fading and Correlation Effects
Analyze the effect for fading signal under the influence of magnitude and impact of
correlation of two signal elements. Marks: 10 CO: 4 RBT Level: Analyze

2.1 Introduction
Fading, caused by multipath propagation, results in signal amplitude and phase
fluctuations. The magnitude of fading affects signal quality, while correlation
between signal elements (e.g., antennas) impacts diversity techniques. This anal- ysis
explores fading types, their effects, and correlation’s role in diversity sys- tems.

2.2 Fading Magnitude

Types of Fading

• Rayleigh Fading: Occurs in non-line-of-sight environments. Envelope:

r p(r) =
e−r2/(2σ2), r≥ 0
σ2
Deep fades (20-30 dB below mean) are common.
• Rician Fading: Occurs with a dominant path. Envelope:
( )
r
p(r) = e−(r2+A2)/(2σ2)I Ar0 2

σ2 σ
where A is the dominant path amplitude, I0 is the modified Bessel function.

• Flat vs. Frequency-Selective: Flat when bandwidth B < Bc (coherence

bandwidth); frequency-selective when B > Bc.
• Fast vs. Slow: Fast when Tc < Ts (symbol duration); slow when Tc > Ts.

Effects

• Signal Power Variation: Deep fades reduce signal strength below receiver
sensitivity.
• Reduced SNR: Lower signal power decreases SNR, increasing BER. For BPSK
in Rayleigh fading:

Pe
where γ¯ is average SNR.
 • Increased BER: Severe fading (e.g., ¯γ = 10 dB) yields higher BER than Rician
fading with high K-factor (K = A2/2σ2).

2.3 Correlation of Signal Elements

Definition

Correlation coefficient:
Cov ρ
= (x, y) σxσy where x, y are signal envelopes.

Impact on Diversity

• Low Correlation (ρ ≈ 0): Independent fading enhances diversity gain. For

two-branch selection combining:
Pout = P (γ1 < γth) · P (γ2 < γth)

• High Correlation (ρ > 0): Similar fading patterns reduce diversity gain: Pout =
1 − [1 − F (γth)]2 + ρ[F (γth)]2

6
2.4 Mitigation Strategies
• Spatial Diversity: Antennas spaced λ/2 apart (e.g., 16.5 cm at 900 MHz) reduce
ρ.

• Polarization Diversity: Cross-polarized antennas lower correlation.

• Frequency Diversity: OFDM subcarriers ensure independent fading.
• Time Diversity: Hybrid ARQ retransmits data at different times.

2.5 Example
For two antennas with P (γ < γth) = 0.1:

• Uncorrelated: Pout = 0.01.

• Correlated (ρ = 0.5): Pout ≈ 0.195.

In 4G MIMO, low correlation improves throughput. Diagram Suggestion: Rayleigh
fading envelope vs. time.

2.6 Conclusion
Fading magnitude degrades SNR and increases BER, with Rayleigh fading caus- ing
deeper fades than Rician. Low correlation enhances diversity gain, reducing outage
probability. Mitigation through spatial, frequency, or time diversity is essential for
robust performance.

Question 3: IS-95 in 800-900 MHz Band

Analyze the IS-95 used in transmission link operated in approximate frequency band
800-900 MHz range. Marks: 10 CO: 5 RBT Level: Analyze

3.1 Introduction
IS-95 (cdmaOne), a 2G standard using Code Division Multiple Access (CDMA),
operates in the 800-900 MHz band (824-849 MHz uplink, 869-894 MHz downlink).
This analysis covers its technical features, performance, and comparison with GSM.

3.2 Technical Features

• Frequency Band: 25 MHz per direction, 1.25 MHz channel bandwidth,
1.2288 Mcps chip rate using DSSS.
• Forward Link: 64 Walsh codes for channel separation (pilot, sync, paging,
traffic), PN sequences for cell identification.
 • Reverse Link: Long PN codes for user separation, power control every 1.25 ms.
• Mechanisms:
– Soft Handoff: Connects to multiple base stations, reducing call drops.
– Rake Receivers: Combine multipath components for improved SNR.
– Vocoder: Variable rates (9.6, 4.8, 2.4, 1.2 kbps).
– Data Rates: 14.4 kbps (IS-95A), 115 kbps (IS-95B).

3.3 Performance in 800-900 MHz

• Propagation: Lower frequencies offer better penetration and diffraction,
enhancing indoor coverage.
• Path Loss: Lower than higher bands (e.g., 1800 MHz), improving range.
• Multipath Handling: Rake receivers exploit delays (1-2 µs) in urban set- tings.

3.4 Advantages
• High Capacity: 20-30 users per 1.25 MHz vs. 8 per 200 kHz in GSM.
• Fading Resistance: Rake receivers and soft handoff mitigate multipath.
• Security: Spreading codes enhance privacy.
• Frequency Reuse: Reuse factor of 1.

3.5 Disadvantages
• Complexity: Requires precise synchronization and power control.
• Cost: Higher deployment costs than GSM.

3.6 Comparison with GSM

• Access Method: IS-95 (CDMA) vs. GSM (TDMA/FDMA).
• Capacity: IS-95 supports more users.
• Complexity: GSM is simpler but less robust.

3.7 Example
A 1.25 MHz IS-95 channel supports 20 users at 9.6 kbps, vs. 8 users in a 200 kHz GSM
channel. Diagram Suggestion: IS-95 channel structure with Walsh codes.

8
3.8 Conclusion
IS-95’s CDMA in the 800-900 MHz band provides excellent capacity and fading
resistance, paving the way for 3G systems like CDMA2000. Its complexity is a
drawback compared to GSM, but its performance is superior.

Question 4: GSM Frame Structure

Evaluate the GSM frame structure of 216.66 frames per second where each user
transmits a burst of data with assigned time slot. Marks: 10 CO: 5 RBT Level:
Analyze

4.1 Introduction
GSM, a 2G standard, uses TDMA with 216.66 frames per second, dividing each frame
into 8 time slots. This analysis evaluates its design, efficiency, limitations, and
comparison with IS-95.

4.2 Frame Structure

• Frame Details:
– Rate: 216.66 frames/s.
– Duration: 4.615 ms.
– Time Slots: 8 per frame, each 0.577 ms, 156.25 bits at 270.833 kbps
(GMSK).
• Normal Burst: 3 tail bits, 57 data bits, 1 stealing flag, 26-bit training se-
quence, 1 stealing flag, 57 data bits, 3 tail bits, 8.25 guard bits.
• Hierarchical Structure:
– 26-Frame Multiframe: 120 ms, 24 TCH, 1 SACCH, 1 idle.
– 51-Frame Multiframe: 235.4 ms, for control channels.

– Superframe: 26 × 51 = 1326 frames (∼ 6.12 s).

– Hyperframe: 2048 superframes (∼ 3.5 hours).

4.3 Advantages
• Multiple Access: 8 users per frame optimize spectrum.
• Synchronization: Precise timing avoids interference.

9
 • Interleaving: Combats burst errors.
• DTX: Saves power during silence.

4.4 Limitations
• Low Data Rate: Max 9.6 kbps per slot.
• Frequency Planning: Requires reuse patterns (e.g., 4/12).
• Latency: Call setup and handoff delays.

4.5 Comparison with IS-95

• Access Method: GSM (TDMA/FDMA) vs. IS-95 (CDMA).
• Capacity: GSM supports fewer users.
• Complexity: GSM is simpler but less robust.

4.6 Efficiency
• Spectral Efficiency: ∼ 1.35 bps/Hz.

• Overhead: ∼ 114 data bits out of 156.25 per slot.

Diagram Suggestion: GSM frame and burst structure.

4.7 Conclusion
GSM’s frame structure is efficient for voice and SMS, enabling global 2G domi- nance.
Its low data rate and spectral efficiency limitations led to GPRS and EDGE, but it is
less robust than IS-95’s CDMA.

10
1.

2.
3.

4.