Systolic implementation of multiplication over GF(2m) is usually very efficient in area-time complexity,but its latency is usually very large.Thus,two low latency systolic multipliers over GF(2m) based on general irre...Systolic implementation of multiplication over GF(2m) is usually very efficient in area-time complexity,but its latency is usually very large.Thus,two low latency systolic multipliers over GF(2m) based on general irreducible polynomials and irreducible pentanomials are presented.First,a signal flow graph(SFG) is used to represent the algorithm for multiplication over GF(2m).Then,the two low latency systolic structures for multiplications over GF(2m) based on general irreducible polynomials and pentanomials are presented from the SFG by suitable cut-set retiming,respectively.Analysis indicates that the proposed two low latency designs involve at least one-third less area-delay product when compared with the existing designs,To the authors' knowledge,the time-complexity of the structures is the lowest found in literature for systolic GF(2m) multipliers based on general irreducible polynomials and pentanomials.The proposed low latency designs are regular and modular,and therefore they are suitable for many time critical applications.展开更多
A finite element reconstruction algorithm for ultrasound tomography based on the Helmholtz equation in frequency domain is presented to monitor the grouting defects in reinforced concrete structures.In this algorithm,...A finite element reconstruction algorithm for ultrasound tomography based on the Helmholtz equation in frequency domain is presented to monitor the grouting defects in reinforced concrete structures.In this algorithm,a hybrid regularizations-based iterative Newton method is implemented to provide stable inverse solutions.Furthermore,a dual mesh scheme and an adjoint method are adopted to reduce the computation cost and improve the efficiency of reconstruction.Simultaneous reconstruction of both acoustic velocity and attenuation coefficient for a reinforced concrete model is achieved with multiple frequency data.The algorithm is evaluated with numerical simulation under various practical scenarios including varied transmission/receiving modes,different noise levels,different source/detector numbers,and different contrast levels between the heterogeneity and background region.Results obtained suggest that the algorithm is insensitive to noise,and the reconstructions are quantitatively accurate in terms of the location,size and acoustic properties of the target over a range of contrast levels.展开更多
基金Project(61174132) supported by the National Natural Science Foundation of ChinaProject(09JJ6098) supported by the Natural Science Foundation of Hunan Province,China
文摘Systolic implementation of multiplication over GF(2m) is usually very efficient in area-time complexity,but its latency is usually very large.Thus,two low latency systolic multipliers over GF(2m) based on general irreducible polynomials and irreducible pentanomials are presented.First,a signal flow graph(SFG) is used to represent the algorithm for multiplication over GF(2m).Then,the two low latency systolic structures for multiplications over GF(2m) based on general irreducible polynomials and pentanomials are presented from the SFG by suitable cut-set retiming,respectively.Analysis indicates that the proposed two low latency designs involve at least one-third less area-delay product when compared with the existing designs,To the authors' knowledge,the time-complexity of the structures is the lowest found in literature for systolic GF(2m) multipliers based on general irreducible polynomials and pentanomials.The proposed low latency designs are regular and modular,and therefore they are suitable for many time critical applications.
基金Project(31200748)supported by the National Natural Science Foundation of China
文摘A finite element reconstruction algorithm for ultrasound tomography based on the Helmholtz equation in frequency domain is presented to monitor the grouting defects in reinforced concrete structures.In this algorithm,a hybrid regularizations-based iterative Newton method is implemented to provide stable inverse solutions.Furthermore,a dual mesh scheme and an adjoint method are adopted to reduce the computation cost and improve the efficiency of reconstruction.Simultaneous reconstruction of both acoustic velocity and attenuation coefficient for a reinforced concrete model is achieved with multiple frequency data.The algorithm is evaluated with numerical simulation under various practical scenarios including varied transmission/receiving modes,different noise levels,different source/detector numbers,and different contrast levels between the heterogeneity and background region.Results obtained suggest that the algorithm is insensitive to noise,and the reconstructions are quantitatively accurate in terms of the location,size and acoustic properties of the target over a range of contrast levels.
