Deep space networks,satellite networks,ad hoc networks,and the Internet can be modeled as DTNs(Delay Tolerant Networks).As a fundamental problem,the maximum flow problem is of vital importance for routing and service ...Deep space networks,satellite networks,ad hoc networks,and the Internet can be modeled as DTNs(Delay Tolerant Networks).As a fundamental problem,the maximum flow problem is of vital importance for routing and service scheduling in networks.However,there exists no permanent end-to-end path since the topology and the characteristics of links are time-variant,resulting in a crucial maximum flow problem in DTNs.In this paper,we focus on the single-source-single-sink maximum flow problem of buffer-limited DTNs,followed by a valid algorithm to solve it.First,the BTAG(Buffer-limited Time Aggregated Graph)is constructed for modeling the buffer-limited DTN.Then,on the basis of BTAG,the two-way cache transfer series and the relevant transfer rules are designed,and thus a BTAG-based maximum flow algorithm is proposed to solve the maximum flow problem in buffer-limited DTNs.Finally,a numerical example is given to demonstrate the effectiveness of the proposed algorithm.展开更多
Most current deep convolutional neural networks can achieve excellent results on a single image superresolution and are trained using corresponding high-resolution(HR)images and low-resolution(LR)images.Conversely,the...Most current deep convolutional neural networks can achieve excellent results on a single image superresolution and are trained using corresponding high-resolution(HR)images and low-resolution(LR)images.Conversely,their superresolution performance in real-world superresolution tests is reduced because thesemethods create paired LR images by simply interpolating and downsampling HR images,which is very different from natural degradation.In this article,we design a new unsupervised framework conditioned by degradation representations of real-world hyperresolution problems.The approach presented in this paper consists of three stages:we first learn the implicit degradation representation from real-world LR images and then acquire LR images by shrinking the network,which will share similar degradation with real-world images.Finally,we make paired data of the generated real LR images and HR images for training the SR network.Our approach can obtain better results than the recent SR approach on the NTIRE2020 real-world SR challenge Track1 dataset.展开更多
基金supported by the National Science Foundation(Nos.91338115,61231008)National S&T Major Project(No.2015ZX03002006)+2 种基金the Fundamental Research Funds for the Central Universities(Nos.WRYB142208,JB140117)Shanghai Aerospace Science and Technology Innovation Fund(No.201454)the 111 Project(No.B08038).
文摘Deep space networks,satellite networks,ad hoc networks,and the Internet can be modeled as DTNs(Delay Tolerant Networks).As a fundamental problem,the maximum flow problem is of vital importance for routing and service scheduling in networks.However,there exists no permanent end-to-end path since the topology and the characteristics of links are time-variant,resulting in a crucial maximum flow problem in DTNs.In this paper,we focus on the single-source-single-sink maximum flow problem of buffer-limited DTNs,followed by a valid algorithm to solve it.First,the BTAG(Buffer-limited Time Aggregated Graph)is constructed for modeling the buffer-limited DTN.Then,on the basis of BTAG,the two-way cache transfer series and the relevant transfer rules are designed,and thus a BTAG-based maximum flow algorithm is proposed to solve the maximum flow problem in buffer-limited DTNs.Finally,a numerical example is given to demonstrate the effectiveness of the proposed algorithm.
基金Support Plan for Core Technology Research and Engineering Verification of Development and Reform Commission of Shenzhen Municipality (number 202100036).
文摘Most current deep convolutional neural networks can achieve excellent results on a single image superresolution and are trained using corresponding high-resolution(HR)images and low-resolution(LR)images.Conversely,their superresolution performance in real-world superresolution tests is reduced because thesemethods create paired LR images by simply interpolating and downsampling HR images,which is very different from natural degradation.In this article,we design a new unsupervised framework conditioned by degradation representations of real-world hyperresolution problems.The approach presented in this paper consists of three stages:we first learn the implicit degradation representation from real-world LR images and then acquire LR images by shrinking the network,which will share similar degradation with real-world images.Finally,we make paired data of the generated real LR images and HR images for training the SR network.Our approach can obtain better results than the recent SR approach on the NTIRE2020 real-world SR challenge Track1 dataset.
