⚡ electrical engineering

Turbo Receiver Design for Differentially Encoded PSK in Bursty Impulsive Noise Channels

This paper proposes and evaluates optimal and suboptimal turbo receiver designs for differentially encoded PSK in bursty impulsive noise channels, demonstrating that incorporating a differential decoder into the MAP-based detector achieves a 4.5 dB performance gain over conventional designs while approaching theoretical information-theoretic bounds.

Original authors: Chin-Hung Chen, Boris Karanov, Wim van Houtom, Yan Wu, Alex Alvarado

Published 2026-03-16

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Chin-Hung Chen, Boris Karanov, Wim van Houtom, Yan Wu, Alex Alvarado

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Picture: Listening to a Radio in a Storm

Imagine you are trying to listen to a radio station (your data) while driving through a massive thunderstorm (the communication channel).

In a perfect world, the radio would just have a little bit of static (Gaussian noise), like a gentle rain. But in the real world, especially near power lines or electric vehicles, the static isn't just rain—it's lightning strikes. These are "impulsive noises." They are sudden, loud, and chaotic bursts that can completely drown out your message.

This paper is about building a super-smart radio receiver that can hear your message clearly even when lightning is striking right next to the antenna.

The Problem: The "Bursty" Noise

Most old-school radios assume noise is like a steady drizzle. They use simple tricks to ignore it, like "clipping" (cutting off loud sounds) or "blanking" (silencing the radio when it gets too loud).

But these tricks are like trying to stop a hurricane by holding your hand up. They don't work well when the noise is bursty.

Bursty Noise: Imagine the lightning doesn't just strike once; it strikes in a cluster. One strike is followed by another, and another, for a few seconds.
The Challenge: If your radio doesn't know that the noise is "clumping" together, it gets confused and throws away good data along with the bad.

The Solution: The "Turbo" Receiver

The authors propose a new receiver design that works like a detective team working together to solve a mystery. They call it a "Turbo Receiver."

Here is how their system works, broken down into three parts:

1. The "Differential" Trick (The Memory Game)

Usually, radios try to guess the exact pitch of a note to understand the message. But if the storm changes the pitch, the radio gets lost.

The Old Way: "What note is this?" (Hard to guess in a storm).
The New Way (Differential): "Is this note higher or lower than the previous one?"
The Analogy: Imagine you are walking through a foggy forest. You don't need to know exactly where you are on the map. You just need to know, "Am I walking uphill or downhill compared to where I was a second ago?" This method is much more robust against the "fog" (noise).

2. The "Hidden Markov" Detective (Predicting the Storm)

The receiver uses a mathematical model called a Hidden Markov Model (HMM).

The Analogy: Imagine you are trying to guess if it's raining outside, but you can't see the sky. You look at the ground.
- If the ground is dry, it's probably sunny.
- If the ground is wet, it might have rained.
- The Secret: If the ground is very wet, it's likely to stay wet for a while (the "burst").
The receiver uses this logic. If it detects a loud noise burst, it predicts, "Okay, the next few seconds will probably be noisy too." It uses this prediction to ignore the noise and focus on the signal.

3. The "Turbo" Team-Up (The Feedback Loop)

This is the "Turbo" part. The receiver has two main detectives:

The Noise Detective: Figures out which parts of the signal are noise.
The Message Decoder: Figures out what the actual message is.

In the past, these detectives worked separately. The Noise Detective would clean the signal and pass it to the Decoder.

The Innovation: In this new design, they work together in a loop. The Decoder says, "Hey, I think this part is a '1', not noise!" and tells the Noise Detective. The Noise Detective says, "Oh, you're right, I'll adjust my guess."
They pass notes back and forth (iterations), getting smarter with every exchange, until they are almost 100% sure of the message.

The Two Versions: The Ferrari and the Hybrid

The paper proposes two versions of this receiver:

The "Optimal Joint" Design (The Ferrari):
- This combines the Noise Detective and the Message Decoder into one giant, super-complex brain.
- Pros: It is the fastest and most accurate. It gets very close to the theoretical limit of how fast you can talk in a storm.
- Cons: It is very expensive to build (high computational complexity). It requires a lot of processing power.
The "Suboptimal Separate" Design (The Hybrid):
- This splits the detectives back into two separate teams, but they still talk to each other.
- Pros: It is half as complex (cheaper and faster to run) as the Ferrari.
- Cons: It loses a tiny bit of performance (about 0.2 dB in the worst cases), but for most real-world situations, it's almost as good as the Ferrari.

Why This Matters (The Results)

The authors ran massive simulations to test their ideas. Here is what they found:

Huge Gains: Compared to old receivers, their new design improved performance by 4.5 dB. In the world of radio, that is a massive jump. It's the difference between hearing a whisper clearly and hearing nothing but static.
Close to Perfect: Their receiver gets within 1 dB of the absolute theoretical limit. It's as good as physics allows.
Robustness: Even if the receiver doesn't know the exact details of the storm (mismatched parameters), it still works very well. It's like a car that drives great even if the GPS is slightly off.

Summary

This paper teaches us how to build a radio receiver that doesn't just "brute force" its way through noise. Instead, it uses memory (to predict bursts), differential coding (to ignore pitch shifts), and teamwork (turbo iterations) to listen clearly through the loudest storms.

They offer a "Ferrari" version for when you need maximum speed and a "Hybrid" version that is 50% cheaper but still drives almost as fast. This is a major step forward for wireless communication in electric vehicles, power grids, and other noisy environments.

1. Problem Statement

Wireless communication systems, particularly those in electric vehicles (EVs) and power substations, face significant challenges from Impulsive Noise (IN) generated by power devices (e.g., switching gears, DC-DC converters).

Limitations of Current Solutions: Conventional mitigation techniques like clipping and blanking are simple but suboptimal, especially in severe IN conditions.
Modeling Gap: While the Middleton Class A model is widely used, it assumes memoryless noise and fails to capture the bursty temporal correlations observed in real-world IN. The Markov-Middleton model addresses this by combining Middleton statistics with Hidden Markov Models (HMMs), but a comprehensive information-theoretic analysis and optimal receiver design for this specific model using complex modulation (PSK) have been lacking.
Modulation Context: Many wireless standards (e.g., DVB-T2, WiFi 802.11g) use Differentially Encoded (DE) PSK to handle phase ambiguity. However, existing robust IN receivers primarily focus on simple BPSK or non-differential schemes, missing the potential gains of integrating differential decoding with IN detection.

2. Methodology

A. Channel Modeling and Information-Theoretic Analysis

The authors utilize the Markov-Middleton model, characterized by:

Parameters: Background noise variance ( $\sigma_0^2$ ), impulsive index ( $A$ ), impulsive-to-background power ratio ( $\Lambda$ ), and correlation parameter ( $r$ ).
HMM Representation: The noise states follow a Markov chain, allowing the modeling of bursty noise behavior.
Achievable Information Rate (AIR): The paper derives the AIR for PSK inputs over this channel using a simulation-based sum-product algorithm with an auxiliary state process. They analyze both matched (perfect channel knowledge) and mismatched (imperfect parameter estimation) decoding scenarios.

B. Receiver Architecture Design

The authors propose a Turbo Receiver structure that integrates three components:

Convolutional Encoder: Rate-1/2 convolutional code.
Differential Encoder: Encodes information in the phase difference between successive symbols (modeled as a rate-1 convolutional code).
MAP-Based Joint Detector/Demapper: A unified Hidden Markov Model (HMM) framework is used to derive the Maximum A-Posteriori (MAP) metrics.

Two specific receiver designs are proposed:

Optimal Joint Turbo-DE-PSK-IN Receiver:
- Constructs a super-trellis where the IN detector and the DE-PSK demapper are combined into a single MAP block.
- Iteratively exchanges extrinsic information between the joint demapper and the convolutional decoder.
- Key Innovation: Incorporates the differential decoder directly into the IN detection process, leveraging the recursive nature of differential encoding to increase time diversity.
Suboptimal Separate Turbo-DE-PSK-IN Receiver:
- Decouples the IN detector from the DE-PSK demapper.
- The IN detector runs once to provide soft outputs, which are then fed into a separate DE-PSK demapper and convolutional decoder.
- Goal: To reduce computational complexity while maintaining near-optimal performance.

C. Complexity Analysis

The authors analyze the computational complexity (number of multiplications in the BCJR algorithm). They demonstrate that while the joint design is optimal, its complexity grows significantly with the number of turbo iterations. The separate design offers a trade-off, reducing complexity by approximately 50% for 10 iterations with minimal performance loss.

3. Key Contributions

Unified HMM Framework: Provided a general representation for deriving AIR, MAP detectors, and iterative decoding metrics, revealing commonalities across different algorithms under a forward-backward framework.
Comprehensive AIR Analysis: Extended AIR analysis to finite-state Markov-Middleton channels with PSK inputs, covering matched and mismatched scenarios. They identified a non-monotonic relationship between AIR and the impulsive-to-background noise ratio ( $\Lambda$ ), explaining it via likelihood function behavior.
Optimal Joint Receiver Design: Proposed a novel turbo receiver that integrates differential decoding with IN detection.
- Achieves a 4.5 dB gain over conventional turbo-PSK-IN receivers (at BER $10^{-4}$ ).
- Operates within ~1 dB of the theoretical AIR bound.
Suboptimal Low-Complexity Design: Proposed a separate receiver design with half the complexity of the joint design, showing negligible performance degradation in mild to moderate IN conditions.
Robustness Validation: Demonstrated that the proposed designs are robust against:
- Mismatched State Models: Using a 2-state model instead of the true 4-state model yields similar performance.
- Parameter Estimation Errors: The system remains effective even with imperfect knowledge of noise parameters ( $A, \Lambda, r$ ).

4. Simulation Results

Performance Gain: In simulations with a 4-state Markov-Middleton channel ( $A=0.1, 0.3, 0.5$ $A = 0.1, 0.3, 0.5$ ; $\Lambda=10$ $Λ = 10$ ; $r=0.9$ $r = 0.9$ ), the proposed Turbo-DE-QPSK-IN receiver significantly outperformed the conventional Turbo-QPSK-IN receiver.
- At $A=0.1$ , the gain was 4.5 dB at BER $10^{-4}$ .
- The performance gap to the theoretical bound was approximately 1.1 to 2.1 dB.
Comparison with LDPC: Despite using a simple convolutional code, the proposed receiver approached the performance of state-of-the-art LDPC-based receivers (as seen in DVB-S2 standards) after 10 iterations.
Interleaver Depth: The system is sensitive to interleaver depth. A depth of 100,000 bits is ideal, while reducing it to 800 bits causes a ~10 dB loss. The differential structure helps spread errors effectively, a benefit not fully realized in non-differential systems.
Complexity vs. Performance: The suboptimal separate design achieved nearly identical BER performance to the joint design for $A=0.1$ and $0.3$, with only a 0.2 dB loss in severe conditions ( $A=0.5$ ), validating its use for practical implementation.

5. Significance

This paper bridges the gap between theoretical information limits and practical receiver design for bursty impulsive noise environments.

Practical Impact: It offers a concrete solution for wireless systems operating in high-interference environments (EVs, power grids) by adapting standard differential modulation schemes to be robust against IN.
Design Guidance: The analysis of mismatched decoding provides engineers with guidelines on how much channel parameter estimation accuracy is required and how to simplify receiver state spaces (e.g., using fewer noise states) without sacrificing performance.
Future Framework: The unified HMM approach presented is extensible to other modulation formats (e.g., QAM) and coding schemes (e.g., LDPC), suggesting a pathway for future high-performance communication systems in noisy industrial environments.