Existing deep unfolding methods unroll an optimization algorithm with a fixed number of steps,and utilize convolutional neural networks(CNNs)to learn data-driven priors.However,their performance is limited for two mai...Existing deep unfolding methods unroll an optimization algorithm with a fixed number of steps,and utilize convolutional neural networks(CNNs)to learn data-driven priors.However,their performance is limited for two main reasons.Firstly,priors learned in deep feature space need to be converted to the image space at each iteration step,which limits the depth of CNNs and prevents CNNs from exploiting contextual information.Secondly,existing methods only learn deep priors at the single full-resolution scale,so ignore the benefits of multi-scale context in dealing with high level noise.To address these issues,we explicitly consider the image denoising process in the deep feature space and propose the deep unfolding multi-scale regularizer network(DUMRN)for image denoising.The core of DUMRN is the feature-based denoising module(FDM)that directly removes noise in the deep feature space.In each FDM,we construct a multi-scale regularizer block to learn deep prior information from multi-resolution features.We build the DUMRN by stacking a sequence of FDMs and train it in an end-to-end manner.Experimental results on synthetic and real-world benchmarks demonstrate that DUMRN performs favorably compared to state-of-theart methods.展开更多
In the 6G environment,addressing challenges like missing data,demodulation errors,and offgrid issues during target parameter estimation is a significant hurdle for integrated sensing and communication(ISAC)systems.In ...In the 6G environment,addressing challenges like missing data,demodulation errors,and offgrid issues during target parameter estimation is a significant hurdle for integrated sensing and communication(ISAC)systems.In the ISAC framework,a commonly used method for parameter estimation is compressive sensing.However,it often struggles with off-grid problems in continuous parameter estimation.In contrast,the atomic norm has been proven effective in overcoming these off-grid issues,making it a more suitable approach for continuous parameter estimation.In this paper,we investigate the application of atomic norm in ISAC and propose an ISAC model based on orthogonal frequency division multiplexing(OFDM)for parameter estimation.We utilize the atomic norm under conditions of incomplete data and demodulation errors.To enhance the convergence speed and accuracy of our algorithm,we implement the alternating direction method of multipliers(ADMM)for iterative processing.We refer to this algorithm as ANMI.Building on this foundation,we develop a deep unfolding network algorithm,ANMIADMM-Net,which further mitigates the impact of missing data and demodulation errors on target parameter estimation by training optimal parameters.Experimental results demonstrate that our proposed ANMI and ANMI-ADMM-Net accurately estimate target parameters even in the presence of missing data and demodulation errors,exhibiting superior precision and robustness compared to traditional methods.展开更多
基金partially supported by the National Key R&D Program of China(No.2020YFA0714101)the National Nature Science Foundation of China(Nos.61872162,62102414,62172415,and 52175493).
文摘Existing deep unfolding methods unroll an optimization algorithm with a fixed number of steps,and utilize convolutional neural networks(CNNs)to learn data-driven priors.However,their performance is limited for two main reasons.Firstly,priors learned in deep feature space need to be converted to the image space at each iteration step,which limits the depth of CNNs and prevents CNNs from exploiting contextual information.Secondly,existing methods only learn deep priors at the single full-resolution scale,so ignore the benefits of multi-scale context in dealing with high level noise.To address these issues,we explicitly consider the image denoising process in the deep feature space and propose the deep unfolding multi-scale regularizer network(DUMRN)for image denoising.The core of DUMRN is the feature-based denoising module(FDM)that directly removes noise in the deep feature space.In each FDM,we construct a multi-scale regularizer block to learn deep prior information from multi-resolution features.We build the DUMRN by stacking a sequence of FDMs and train it in an end-to-end manner.Experimental results on synthetic and real-world benchmarks demonstrate that DUMRN performs favorably compared to state-of-theart methods.
基金supported in part by Natural Science Foundation of China under Grant 62571133 and 62571135Natural Science Foundation of Fujian Province under Grant 2025J01459.
文摘In the 6G environment,addressing challenges like missing data,demodulation errors,and offgrid issues during target parameter estimation is a significant hurdle for integrated sensing and communication(ISAC)systems.In the ISAC framework,a commonly used method for parameter estimation is compressive sensing.However,it often struggles with off-grid problems in continuous parameter estimation.In contrast,the atomic norm has been proven effective in overcoming these off-grid issues,making it a more suitable approach for continuous parameter estimation.In this paper,we investigate the application of atomic norm in ISAC and propose an ISAC model based on orthogonal frequency division multiplexing(OFDM)for parameter estimation.We utilize the atomic norm under conditions of incomplete data and demodulation errors.To enhance the convergence speed and accuracy of our algorithm,we implement the alternating direction method of multipliers(ADMM)for iterative processing.We refer to this algorithm as ANMI.Building on this foundation,we develop a deep unfolding network algorithm,ANMIADMM-Net,which further mitigates the impact of missing data and demodulation errors on target parameter estimation by training optimal parameters.Experimental results demonstrate that our proposed ANMI and ANMI-ADMM-Net accurately estimate target parameters even in the presence of missing data and demodulation errors,exhibiting superior precision and robustness compared to traditional methods.
