Robust PSD Map Construction via Sparse Regression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current wireless communication systems face challenges in coordinating the use of radio frequency spectrum among different technologies, leading to interference issues, especially in accurately localizing primary users and estimating power spectral density (PSD) across space and frequency, due to model uncertainties and sparse presence of active users.
Innovation Solution
A collaborative sensing scheme using a parsimonious system model and basis expansion to approximate PSD distribution, employing group sparse regression and alternating direction method of multipliers (ADMoM) for efficient PSD map construction, while accounting for hierarchical sparsity and channel uncertainties through robust group sparse least-absolute-shrinkage-and-selection operator (Lasso) and total least-squares techniques.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a grid-based approach is used for PU transmitter locations, then spatial-domain sparsity emerges enabling efficient PSD estimation, but model offsets occur as actual PU locations may not coincide with grid points
Solution Approach 1:
The patent performs preliminary channel cartography to acquire grid-to-CR channel gains before the actual PSD estimation process. This preliminary action creates a robust regression matrix that accounts for spatial location uncertainties, allowing the subsequent sparse regression to proceed with reduced sensitivity to grid-mismatch errors.
Solution Approach 2:
The patent transforms the PSD estimation problem from a direct spectral analysis to a sparse regression problem in the spatial domain. By changing the parameter representation from frequency-domain PSD values to spatial-domain basis coefficients, the method exploits sparsity to improve estimation accuracy while using channel cartography to compensate for spatial modeling errors.
2Measurement precision
If channel gain cartography is used to acquire regression matrix information, then sensing performance improves, but inaccurate channel gains or shadowing-agnostic path loss models deteriorate sensing accuracy
Solution Approach 1:
The patent performs preliminary channel cartography to acquire grid-to-CR channel gains before the actual PSD estimation process. This preliminary action creates a robust regression matrix that accounts for spatial location uncertainties, allowing the subsequent sparse regression to proceed with reduced sensitivity to grid-mismatch errors.
Solution Approach 2:
The patent introduces channel cartography as an intermediary process that measures and stores channel gains between grid points and cognitive radios. This intermediary measurement layer decouples the PSD estimation accuracy from the accuracy of path loss models, as the actual measured channel gains are used in the regression matrix rather than theoretical predictions.
3Productivity
If sparse regression is used for PSD estimation, then computational efficiency improves, but model uncertainties and outliers reduce estimation reliability
Solution Approach 1:
The patent performs preliminary channel cartography to acquire grid-to-CR channel gains before the actual PSD estimation process. This preliminary action creates a robust regression matrix that accounts for spatial location uncertainties, allowing the subsequent sparse regression to proceed with reduced sensitivity to grid-mismatch errors.
Solution Approach 2:
The patent transforms the PSD estimation problem from a direct spectral analysis to a sparse regression problem in the spatial domain. By changing the parameter representation from frequency-domain PSD values to spatial-domain basis coefficients, the method exploits sparsity to improve estimation accuracy while using channel cartography to compensate for spatial modeling errors.
Data Source
AI summary
This disclosure describes techniques for constructing power spectral density (PSD) maps representative of the distribution of radio frequency (RF) power as a function of both frequency and space (geographic location). For example, the disclosure describes techniques for construction PSD maps using robust basis pursuit forms of signal expansion.