Receiver Parameter Tuning for Lower BER Under Non-Gaussian Noise
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for selecting system parameters, such as full gradient descent and cost functions assuming Gaussian noise, are computationally intensive and limited in their applicability to non-Gaussian noise scenarios, often resulting in suboptimal Bit Error Rate (BER) performance in systems like hard disk read channels.
Innovation Solution
A stochastic gradient descent technique using a modified cost function that incorporates near error events and a relaxed unit step function to update system parameters, allowing for more efficient and general parameter selection across various noise conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If full gradient descent technique is used to adjust system parameters, then parameter optimization is achieved, but computational complexity and time consumption increase significantly
Solution Approach 1:
The patent extracts and addresses only the most significant error events and near-error events rather than processing all possible variations. By focusing computational resources on the most impactful events (those with small metric differences), the system achieves effective parameter optimization without the exhaustive computational burden of full gradient descent.
Solution Approach 2:
The patent applies partial action by considering only a subset of error events (those meeting specific criteria for significance) rather than all possible events. This selective approach processes enough events to achieve convergence while avoiding the excessive computation of examining every possible parameter variation.
2Productivity
If Gaussian noise assumption is used in cost function formulation, then computational speed is improved, but accuracy deteriorates when non-Gaussian noise is present
Solution Approach 1:
The patent changes the parameter assumptions in the cost function from Gaussian noise to a more general framework that accommodates non-Gaussian noise characteristics. This is achieved by modifying how error events are weighted and processed, allowing the system to adapt to actual noise conditions in the channel while maintaining computational efficiency through the selective event processing approach.
3Ease of manufacture
If conventional parameter selection methods are used, then manufacturing process is simple, but adaptability to different noise conditions is poor
Solution Approach 1:
The patent implements self-service by enabling the system to automatically adapt its parameters based on observed error events and near-error events during operation. Rather than requiring manual reconfiguration for different noise conditions, the system autonomously processes metric differences from actual channel performance and adjusts parameters accordingly, achieving both ease of manufacture and adaptability.
Data Source
AI summary
A set of one or more receiver parameters is adjusted. It is determined whether to adjust the set of receiver parameters. In the event it is determined to adjust the set of receiver parameters, a new set of values is generated for the set of receiver parameters using a cost function (where the cost function does not assume a noise signal in a receive signal to have a particular statistical distribution) and the set of receiver parameters is changed to have the new set of values.