number of right-hand sides to be solved for the complete monostatic RCS calculation is 91 for each example. It can be seen that the number of MVPs of the SAI-GMRES method varies largely with respect to incident angles (i.e., right-hand sides), while that of the GMRESE methods are more similar and almost constant for each example. This is another advantage of the GMRESE method proposed in this article. More detailed comparisons can be found in Table 5, in which the cumulated number of matrix-vector products and the total elapsed solution time for a complete monostatic RCS calculation are summarized using both the SAI-GMRES method and the GMRESE methods. In this table, it can be observed that the overall gain ranges from a factor of 3.5 to 5.9 when compared in terms of CPU time.
CONCLUSIONS AND COMMENTSIn this article, we have described an enhanced GMRES method combined with MLFMA for solving linear systems from electromagnetic wave scattering problems. We explicitly compute and save the approximate eigenvectors associated with the eigenvalues close to the origin, either by the ARPACK software in a preprocessing phase or by the GMRES-DR method as a byproduct during the iteration process. The spectral information is then augmented into the Krylov subspace of the GMRES method to recover global information in the MLFMA context. Numerical experiments of typical examples are performed and comparison is made with the SAI preconditioned GMRES method, which shows the GMRESE method proposed in this article is more efficient and can reduce significantly both the iteration number and the overall simulation time.The novelty of this approach presented in this article compared to the augmentation techniques in Ref. 9 is that the proposed method induces a global deflation in the spectrum of the preconditioned matrix. The combined effect of the preconditioning techniques that cluster most of the eigenvalues to one, coupled with the augmentation technique that shifts the rest of the smallest eigenvalues of the spectrum can be very beneficial for the convergence of the Krylov iterative methods. Moreover, the price of augmenting eigenvectors into the Krylov subspace in the classical augmented GMRES method, that is the k additional matrix vector products, is avoided, since the spectral information is explicitly saved and applied at every cycle. Recently, radio frequency identification (RFID) in the UHF band has gained popularity in many applications, because it provides a long reading distance, fast reading speed, and large information storage capability. In RFID systems, tags are usually attached to objects having various material properties. Among them, metallic objects strongly affect the performance of a tag antenna including radiation efficiency and gain levels. To overcome these problems, planar inverted-F and microstrip patch antennas have been proposed for RFID tag applications. However, these antennas have a narrow bandwidth and thick height, and the resonant frequency can easily be shifted depending on the characteristics o...
Fully integrated 0.18‐µm CMOS dual‐band band‐pass filtering transmit/receive (T/R) switch operating in two frequency bands of 35.5–43.7 and 56.5–63 GHz is reported. The developed T/R switch consists of three single‐pole double‐throw switches, each designed based on a band‐pass filter with shunt nMOS transistors performing the switching function. The measured insertion losses of the switch are 8.9/12.5 and 10/12.7 dB for receiving and transmitting operations at 40/60 GHz, respectively. The measured stop‐band rejection ratio between 40 and 51 GHz is over 30/22 dB for two operations, respectively. The measured isolation is 56/51 and 57/51 dB for two operations at 40/60 GHz, respectively. The measured input 1‐dB compression points of the switch are 23, 16.5 for single‐tone 40‐, 60‐GHz input signal and 18/14 dBm at 40/60 GHz for concurrent dual‐tone 40/60‐GHz input signals for transmitting operation, respectively. The total chip size is 1840 µm × 860 µm excluding all the testing pads.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.