By extending the space-alternating generalised expectation-maximisation (SAGE) parameter estimation algorithm to include a novel cost variant process, the robustness of direction of arrival (DoA) estimation from the SAGE algorithm is shown to improve. A novel extension to the standard SAGE algorithm that enhances the parameter estimation process is first introduced. Field trials that have been conducted using an adaptive antenna to verify the performance of this algorithm are presented. Results show that the extended SAGE algorithm improves DoA performance compared to standard SAGE by 5 degrees, when operating in an urban environment.
An algorithm is presented for efficient computation of electromagnetic interactions between a large number of sources in electrically small problems. The algorithm is based on a plane-wave expansion of the free-space Green's function. The expansion consists of both propagating and evanescent plane waves, and is stable at low frequencies. The algorithm is used in the iterative solution procedure of the method of moments to reduce the computational complexity of solving the matrix equation. It reduces the complexity of the matrix-vector multiplication from O(N2) to O(N log N). The numerical results verify the validity and efficiency of the algorithm in solving large-scale and low-frequency problems.
The phase noise of an oscillator with a thin-film barium strontium titanate (BST) capacitive tuning element, or varactor, is characterised and benchmarked against the same oscillator with a silicon semiconductor junction varactor. Phase noise tracks closely with varactor Q within a specific voltage range as expected. Compared to the semiconductor varactor-based oscillator, the BST-based oscillator demonstrates reduced phase noise degradation near zero volts, but greater phase noise degradation when operated near breakdown.
Spherical lens antennas can produce multiple beams from multiple feeds, and these may be scanned to any angle. While an electrically large, uniform dielectric sphere offers mediocre aperture efficiency, this can be usefully increased by varying the dielectric constant with radius, as in the Luneburg lens. A continuous radial variation is difficult to achieve in practice and so a series of concentric shells is often preferred. A useful variant is that of the hemisphere lens used with a ground plane: this offers a relatively low-profile solution which is advantageous for a vehicle-mounted scanning antenna. A theory for the hemisphere lens radiation pattern is developed where this is described as the superposition of two spherical lens patterns i.e. that of the real and virtual sources. Measured radiation patterns of such lens antennas are reported which validate this theory. A prototype lens antenna is then reported which uses only two layers of dielectric material, which presents significant fabrication cost savings compared to a traditional Luneburg antenna. This 236 turn diameter lens offers 35.1 dBi of gain at 28 GHz (68% aperture efficiency) while scanning over a 150 degrees solid angle.
Bandstop filters (BSFs) with high skirt selectivity as well as wide rejection bandwidth are designed by means of open-ended lambda g/4 stepped-impedance resonators (SIRs) and spur-line sections. Base on the formula of an SIR, the frequency of a transmission zero (fz) is accurately predicted. It is found that the impedance and length ratio can be simultaneously obtained to improve stopband performance of the filters. The transmission zero is purposely located at a lower frequency to achieve the high skirt selectivity. On the contrary, if it is adequately situated at a higher frequency, the stopband bandwidth can be increased. Also, the spur-line structure is investigated by transmission line theory and applied to provide an extra transmission zero in the stopband. With a proper choice of the even- and odd-mode characteristic impedances of the spur-line, the stopband bandwidth can be further extended. Realised by planar microstrip technology, the proposed BSF with its high skirt selectivity has a maximal value of 86.8 dB/GHz, while the other type of BSF can further widen the -20 dB stopband by 29% significantly, when incorporating the spur-line into the BSF with an SIR. Both have a size reduction benefit of better than 40% with respective to the conventional case
Microstrip patch antennas mounted over a high impedance electromagnetic bandgap (EBG) substrate are studied. The structure is equivalent to a new microstrip antenna, where the conducting ground plane is replaced by a high impedance EBG layer. Initially, the bandgap of the EBG structure is determined. Then, patch antennas with this EBG ground plane are designed to work within and outside the bandgaps. Parametric studies are conducted to maximise their impedance bandwidths and gains. It is found that very wide bandwidths, of around 25%, can be obtained by variation of the original antenna and EBG parameters. Their gains are similarly increased. Sample antennas are also fabricated and tested, to verify the designs.
Quantitative expressions on the locking performance of a MESFET injection-locked oscillator (ILO) array are presented using nodal-admittance matrix representation. The formulations are derived on a linear approach, hence they are effective in the ILO array design. The locking performance of the ILO array includes the free-running frequency, output power and locking bandwidth. The derived formulations are verified using the measured results from a MESFET ILO and a four-element ring-type coupled ILO array.
A study on the feasibility of the application of a Goubau line for millimetre and submillimetre wave sensors is presented. Simulations and calculations are performed for different line dimensions. Sensitivity of gas detection is examined in detail. Coupling effects between Goubau lines are also studied.
Electromagnetic inverse problems have been found in numerous research areas, such as medical imaging, non-destructive testing, geophysical prospecting and microwave tomography. These inverse problems can be considered as a large-scale nonlinear programming problem. In the paper, a quasi-Newton algorithm based on Broyden-Fletcher-Goldfarb-Shanno (BFGS) Hessian matrix updating formula is introduced and applied to solve two-dimensional (2-D) electromagnetic inverse problems. Numerical results of image reconstruction from the simulation data for some dielectric cylinders are given, and the efficiency of the proposed method for image reconstruction is evaluated.
The use of conformal antennas for vehicle applications is growing very rapidly due to the development of modern wireless communication technology and service. One of the popular ways to design automobile conformal antennas is to modify an existing antenna. These modification processes rely on the designer's intuition and many tedious measurements. The authors propose a computational conformal antenna design based on the nondominated sorting genetic algorithm (NSGA) modifying existing antenna types to improve performance. Simulations and measurement results are presented and discussed.
The authors propose a dual-polarised T/R antenna system that utilises a 3 dB 90° hybrid coupler and a dual-polarised antenna. From the practical point of view the T/R isolation performance of this system can be made better than that of systems using circulators. There is no power loss both in transmit and receive modes compared to 6 dB total loss in a conventional system using a hybrid coupler and a single polarised antenna. To validate the system performance, a sample structure is realised for a C band application. The presented system analysis and measurements are for an above-ground application. The measurements show good agreement with the theoretical expectations.
A rectangular spiral antenna with switches is proposed for beam adaptive applications. The excitation of switches introduces variations in the current distribution on the antenna arm, thus causing the beam to steer. A switch is implemented as an element to short-circuit at a point on the spiral arm to the ground conductor (shorted spiral antenna). A shorted spiral configuration using four switches is analysed. For this configuration, both single- and multipoint switching are implemented. Sixteen possible switching cases are investigated. The radiation patterns are measured in the Satimo antenna test facility and numerically supported using the finite-difference time-domain (FDTD) method. Tilted (24°<θmax<44°) and axial (5°<θmax<12°) beams are obtained using various switching cases, thus realising an adaptive antenna. The gain, which is approximately 6.5 dBi, stays uniform within ±1 dB variation and the VSWR remains within an acceptable limit of 2 for the majority of the switching cases.
A perfectly matched layer (PML) formulation is developed for 3-D envelope finite element (EVFE) solvers. The PML performance is tested against different numbers of layers and different sigma(max) values. The results show that the PML can provide sufficient absorption of incident waves when proper parameters are chosen. Numerical tests also show that this 3-D EVFE algorithm with PML boundary conditions is unconditionally stable as the time intervals increase. Finally, the 3-D EVFE solver equipped with PML boundary conditions has been applied to the modelling of microwave integrated circuits such as MMIC interconnects and on-chip parasitic structures. Good agreements with existing simulation results are obtained with improved computational efficiency.
A novel microwave probing application using elastomer mesh is described. It is targeted for testing wafer level packages with very fine pitch of the order of 100 micron and large pin counts of the order of a thousand. The metallised elastomer mesh provides mechanical compliance as well as good electrical contact. A mesh-coplanar probe is modelled by the partial element equivalent circuit (PEEC) method. The model is verified through frequency domain measurements on a prototype test fixture.
The authors present a novel push-pull amplifying array using quadruple antenna-patch couplers and dual-feed antennas. It exploits the advantages of both class-B push-pull amplifiers and active integrated antennas (AIAs), resulting in a high-efficiency, linear and yet compact design. A state-of-the-art heterojunction FET power amplifier of 50% peak power-added efficiency (PAE) is achieved at 10 GHz. A three-element array prototype was successfully built, achieving a peak antenna gain of 19.6 dBi. This is 5.7 and 10.4 dB better than amplifying arrays using the antenna-patch coupler approach and the conventional parallel feeding network, respectively.
‘Frequency scaling’ relates to the variation of propagation effects as a function of frequency. The frequency scaling ratio between the Ka and V bands, derived using a model based on an integrated parameter such as rain rate, is inadequate. In fact, propagation variations due to the microphysical characteristics of rain need to be considered in order to improve the existing scaling model, particularly at higher frequencies. Many different laws for raindrop size distribution, combined with one year of measured drop size distribution (DSD), are used to study propagation sensitivity to drop size and shape. A statistical study of the normalised DSD has been carried out, and two scaling models are proposed in which microphysical considerations are taken into account. Italsat and Olympus data have been used to validate this approach.
Novel low loss 3 x 3-way and 5 x 5-way phase combiners for power amplifier load balancing in wireless network cells are presented. Based on an air-filled coaxial multiring structure, the designs combine very low insertion loss with compact physical profile. The concept is studied for the PCS (personal Communication service) transmit band in the range 1.91-1.99 GHz by simulation. Measurements validate the predicted very low insertion loss for the three-way combiner of 0.05 dB, while return loss and port isolation are both better than 25 dB. Simulations exhibit, with 0.15 dB insertion loss, a similar performance for the five-way device. Return loss and port isolation are better than 23 dB in this case.
A method to extend the range of validity of heuristic diffraction coefficients for an impedance wedge to all diffraction and incidence angles is suggested. The proposed diffraction coefficients enforce continuity of the total field, are reciprocal and symmetrical, and satisfy plausible boundary conditions on the wedge faces. Good agreement with Maliuzhinets' method is observed in all conditions, excluding those for which the first-order impedance boundary conditions are known to be inaccurate.
The paper presents a microstrip patch array antenna for transmitting (Tx) and receiving (Rx) in the Ku band. The patch array antenna has horizontal polarisation for the Rx band and vertical polarisation for the Tx band. The element of the patch array antenna was designed as a three-stacked structure consisting of one radiation patch and two parasitic patches for high gain and wide bandwidth characteristics. The unit elements were arranged in a 1 x 8 array using a mixture of series and parallel feeds. To verify the practicality of this antenna, a three-stacked patch array antenna was fabricated and its performance was measured. The gain of the array antenna in the Rx and Tx bands exceeded 17 and 18dBi, respectively. The impedance bandwidth was over 10% in both bands. The cross-polarisation level was below -25dB, and the sidelobe level was below -9.4 dB.
Channel characterisation is an essential step to the development of outdoor or indoor wireless networks. Indeed, for multimedia applications, new radio mobile systems must accurately take into account channel behaviour. An efficient 2-D ray-tracing method is proposed to characterise the narrowband and wideband radio channels for a very large number of receivers in microcellular configurations. It is based on a quick precalculation of an exact 2-D visibility graph. The proposed method follows an ITU recommendation, which advocates, for wideband characterisation, consideration only of the paths included in a 18-dB dynamic range of power impulse response. Contrary to the classical approach, which consists of thresholding the complete impulse response in a post-treatment, this method only computes the significant paths. The interest of the proposed method resides in its significant computation time reduction factor, in comparison with the classical approach, without any significant loss in accuracy. Received power and wideband parameter maps are computed for about 40000 receivers in a dense urban environment, and are provided with an approximate reduction factor of four and 80% of the null estimation error in comparison to a classical approach.
