Bayesian regression filter and the issue of priors

We propose a Bayesian framework for regression problems, which covers areas which are usually dealt with by function approximation. An online learning algorithm is derived which solves regression problems with a Kalman filter. Its solution always improves with increasing model complexity, without the risk of over-fitting. In the infinite dimension limit it approaches the true Bayesian posterior. The issues of prior selection and over-fitting are also discussed, showing that some of the commonly held beliefs are misleading. The practical implementation is summarised. Simulations using 13 popular publicly available data sets are used to demonstrate the method and highlight important issues concerning the choice of priors.

Simultaneous interrogation of fiber Bragg grating sensors using an acoustooptic tunable filter

We describe a novel technique to provide demultiplexing of fiber Bragg grating sensors, interrogated using interferometric wavelength shift detection. Amplitude modulation of multiple radio frequency driving signals allows an acoustooptic tunable filter to provide wavelength demultiplexing. We demonstrated a noise limited strain resolution of 150 nanostrain/v(Hz) and a crosstalk better than -50 dB.

Fast and widely tunable Bragg grating reflection filter

A fiber Bragg grating filter device linearly tunable over 45 nm is presented. The device has a maximum tuning speed of 19 nm/ms with a wavelength setting time below 1.5 ms.

Portfolio diversification and filter rule profits

Previously, it has been shown that the profits from a simple market timing trading rule applied to a portfolio of shares can be affected by the inter-relationships between the returns of the component securities. In this short letter, the results from applying a more sophisticated 'filter' rule to the same data are reported. Unlike the simple trading rule, the filter rule does produce some evidence of economic profits.

Pulse tail suppression in laser diode output by tunable notch filter

Efficient suppression of relaxation oscillations in the output signal from an overdriven gain-switched laser diode was demonstrated. Several quantum-well distributed feedback laser diodes from different manufacturers were used for experimental analysis. A five-fold increase in the peak power was achieved for the tail-free operation. It was found that spectral filtering removed the nonlinearly chirped components resulting in pulse shortening by a factor of three.

Bidirectional nonreciprocal wavelength-interleaving coherent fiber transversal filter

A bidirectional nonreciprocal wavelength-interleaving filter based on an optically coherent high birefringence fiber transversal filter structure is demonstrated. Stable, low loss, low dispersion, and high isolation operation is demonstrated with reconfigurable transfer characteristics for interleaved channel spacing of 0.8 nm.

Bragg Grating fast Tunable Filter For Wavelength Division Multiplexing

A Bragg grating fast tunable filter prototype working over a linear tuning range of 45 nm with a maximum tuning speed of 21 nm/ms has been realized. The tunable filter system is based on two piezoelectric stack actuators moving a mechanical device thus compressing an apodized fiber Bragg grating. The filter allows both traction and compression and can work in transmission and in reflection. It is designed to work with a channel spacing of 100 GHz according to the ITU specifications for wavelength division multiplexing systems

The ecology of percolating filters containing a plastic filter medium in relation to their efficiency in The treatment of domestic sewage

The suitability of a new plastic supporting medium for biofiltration was tested over a three year period. Tests were carried out on the stability, surface properties, mechanical strength, and dimensions of the medium. There was no evidence to suggest that the medium was deficient in any of these respects. The specific surface (320m2m-3) and the voidage (94%) of the new medium are unlike any other used in bio-filtration and a pilot plant containing two filters was built to observe its effects on ecology and performance. Performance was estimated by chemical analysis and ecology studied by film examination and fauna counts. A system of removable sampling baskets was designed to enable samples to be obtained from two intermediate depths of filter. One of the major operating problems of percolating filters is excessive accumulation of film. The amount of film is influenced by hydraulic and organic load and each filter was run at a different loading. One was operated at 1.2m3m-3day-1 (DOD load 0.24kgm-3day-1) judged at the time to be the lowest filtration rate to offer advantages over conventional media. The other filter was operated at more than twice this loading (2.4m3m-3day-lBOD load 0.55kgm-3day-1) giving a roughly 2.5x and 6x the conventional loadings recommended for a Royal Commission effluent. The amount of film in each filter was normally low (0.05-3kgm(3 as volatile solids) and did not affect efficiency. The evidence collected during the study indicated that the ecology of the filters was normal when compared with the data obtained from the literature relating to filters with mineral media. There were indications that full ecological stability was yet to be reached and this was affecting the efficiency of the filters. The lower rate filter produced an average 87% BOD removal giving a consistent Royal Commission effluent during the summer months. The higher rate filter produced a mean 83% BOD removal but at no stage a consistent Royal Commission effluent. From the data on ecology and performance the filters resembled conventional filters rather than high rate filters.

Fibre Bragg grating sensor interrogation using an acousto-optic tunable filter and low-coherence interferometry

We describe how an acousto-optic tunable filter can be used to both demultiplex the signals from multiple fibre Bragg grating sensors and simultaneously provide wide bandwidth signal demodulation in a system using interferometric wavelength shift detection. In an experimental demonstration, the approach provided a noise limited strain resolution of 24.9 n epsilon Hz(-1/ 2) at 15 Hz.

Development of an electrically tuneable Bragg grating filter in polymer optical fibre operating at 1.55 µm

We present a thorough study on the development of a polymer optical fibre-based tuneable filter utilizing an intra-core Bragg grating that is electrically tuneable, operating at 1.55 µm. The Bragg grating is made tuneable using a thin-film resistive heater deposited on the surface of the fibre. The polymer fibre was coated via the photochemical deposition of a Pd/Cu metallic layer with the procedure induced by VUV radiation at room temperature. The resulting device, when wavelength tuned via Joule heating, underwent a wavelength shift of 2 nm for a moderate input power of 160 mW, a wavelength to input power coefficient of -13.4 pm mW-1 and time constant of 1.7 s-1. A basic theoretical study verified that for this fibre type one can treat the device as a one-dimensional system. The model was extended to include the effect of input electrical power changes on the refractive index of the fibre and subsequently to changes in the Bragg wavelength of the grating, showing excellent agreement with the experimental measurements.

Remotely tuneable optical filter based on polymer fibre Bragg grating

We propose a remotely tuneable optical Bragg grating filter written in polymer optical fibre (POF). Fibre optical pumping in the fibre's absorption bands increases the fibre temperature, which causes a negative wavelength change of the POF Bragg grating. By choosing a proper pumping wavelength remote tuning of the optical filter can be readily realized without changing the gain of the optical signal.

Design of a flat-top fiber Bragg filter via quasi-random modulation of the refractive index

The statistics of the reflection spectrum of a short-correlated disordered fiber Bragg grating are studied. The averaged spectrum appears to be flat inside the bandgap and has significantly suppressed sidelobes compared to the uniform grating of the same bandwidth. This is due to the Anderson localization of the modes of a disordered grating. This observation prompts a new algorithm for designing passband reflection gratings. Using the stochastic invariant imbedding approach it is possible to obtain the probability distribution function for the random reflection coefficient inside the bandgap and obtain both the variance of the averaged reflectivity as well as the distribution of the time delay of the grating.

Iterative subspace analysis based on feature line distance

Nearest feature line-based subspace analysis is first proposed in this paper. Compared with conventional methods, the newly proposed one brings better generalization performance and incremental analysis. The projection point and feature line distance are expressed as a function of a subspace, which is obtained by minimizing the mean square feature line distance. Moreover, by adopting stochastic approximation rule to minimize the objective function in a gradient manner, the new method can be performed in an incremental mode, which makes it working well upon future data. Experimental results on the FERET face database and the UCI satellite image database demonstrate the effectiveness.