Abstract
To account for variations in the frequency, time, and space dimensions, dynamic re-use of licensed bands under the cognitive radio (CR) paradigm calls for innovative network-level sensing algorithms for multi-dimensional spectrum opportunity awareness. Toward this direction, the present paper develops a collaborative scheme whereby CRs cooperate to localize active primary user (PU) transmitters and reconstruct a power spectral density (PSD) map portraying the spatial distribution of power across the monitored area per frequency band and channel coherence interval. The sensing scheme is based on a parsimonious model that accounts for two forms of sparsity: one due to the narrow-band nature of transmit-PSDs compared to the large portion of spectrum that a CR can sense, and another one emerging when adopting a spatial grid of candidate PU locations. Capitalizing on this dual sparsity, an estimator of the model coefficients is obtained based on the group sparse least-absolute-shrinkage-and-selection operator (GS-Lasso). A novel reduced-complexity GS-Lasso solver is developed by resorting to the alternating direction method of multipliers (ADMoM). Robust versions of this GS-Lasso estimator are also introduced using a GS total least-squares (TLS) approach to cope with both uncertainty in the regression matrices, arising due to inaccurate channel estimation and grid-mismatch effects, and unexpected model outliers. In spite of the non-convexity of the GS-TLS criterion, the novel robust algorithm has guaranteed convergence to (at least) a local optimum. The analytical findings are corroborated by numerical tests.
Original language | English (US) |
---|---|
Pages (from-to) | 161-172 |
Number of pages | 12 |
Journal | Physical Communication |
Volume | 5 |
Issue number | 2 |
DOIs | |
State | Published - Jun 2012 |
Bibliographical note
Funding Information:This work was supported by NSF grants CCF-0830480 , CCF-1016605 , ECCS-0824007 and ECCS-1002180 , and QNRF grant NPRP 09-341-2-128 .
Keywords
- Outliers
- Sparse linear regression
- Spectrum cartography
- Spectrum sensing
- Total least-squares