Use Of Nonlinear Filter Research Articles

DeepVehicleSense: An Energy-Efficient Transportation Mode Recognition Leveraging Staged Deep Learning Over Sound Samples

In this paper, we present a new transportation mode recognition system for smartphones called DeepVehicleSense, which is widely applicable to mobile context-aware services. DeepVehicleSense aims at achieving three performance objectives: high accuracy, low latency, and low power consumption at once by exploiting sound characteristics captured from the built-in microphone while being on candidate transportations. To attain high energy efficiency, DeepVehicleSense adopts hierarchical accelerometer-based triggers that minimize the activation of the microphone of smartphones. Further, to achieve high accuracy and low latency, DeepVehicleSense makes use of non-linear filters that can best extract the transportation sound samples. For recognition of five different transportation modes, we design a deep learning based sound classifier using a novel deep neural network architecture with multiple branches. Our staged inference technique can significantly reduce runtime and energy consumption while maintaining high accuracy for the majority of samples. Through 263-hour datasets collected by seven different Android phone models, we demonstrate that DeepVehicleSense achieves the recognition accuracy of 97.44\% with only sound samples of 2 seconds at the power consumption of 35.08 mW on average for all-day monitoring.

Comparative Analysis of Low Earth Orbit Satellite Attitude Determination using Sensor Integration and Attitude Estimation using Kalman Filter

The objective of the research is to undergo a performance comparison in terms of accuracy, convergence time, amount of memory, etc. between the satellite attitude determination and attitude estimation using non-linear filters. The fundamental approach towards it is to design an OBC (On Board Computer) that would help in achieving a controlled output for the chosen plant (MicroMAS1 satellite). The attitude determination algorithm is implemented through TRIAD algorithm, which takes sensor readings of body frame and inertial frame of reference. Then it is used to determine the rotation matrix (DCM) by converting the matrix form into vector form and again back to matrix form to determine the 3x3 matrix, which includes all the Euler angle equations to determine the pitch, roll, yaw characteristics of the system. The attitude estimation algorithms involves the use of nonlinear filters which provide an added advantage that energy can be transferred in a designed manner and extra degrees of freedom are available in filter design. The Unscented Kalman Filter (UKF) is preferred as it addresses the problem using a deterministic sampling approach. Moreover, the non-linear filters are used to remove the noise error and disturbance caused by engine. The design of satellite attitude models involves more of a mathematical approach that would be dealt with MATLAB and SIMULINK operations.

A Predictive Technique for the Real-Time Trajectory Scaling Under High-Order Constraints

Modern robotic systems must be able to react to unexpected environmental events. To this purpose, planning techniques for the real-time generation/modification of trajectories have been developed in recent times. In the frequent case of applications which require following a predefined path, the assigned timing law must be inspected in real time so as to verify whether it satisfies the system constraints, or conversely, if it must be scaled in order to obtain a feasible trajectory. The problem has been addressed in several ways in the literature. One of the known approaches, based on the use of nonlinear filters, is revised in this article in order to return feasible solutions under any circumstances. Differently from alternative strategies, it manages constraints up to the torque derivatives and has evaluation times compatible with the ones required by modern control systems. The proposed technique is validated through simulations and real experiments. Comparisons are proposed with an algorithm based on a model predictive technique and with an alternative scaling system.

Extended Kalman Filter in 2S1 Tracked Vehicle System with Hybrid Control

The essence of the undertaken topic is the problem of estimation of state vector in the model of 2S1 tracked vehicle suspension system through the use of Extended Kalman Filter. The use of non-linear filter has become necessary due to the magnetorheological damper located at suspension system, which has been described by hyperbolic model. Application of the damper caused the tested suspension system has become a semi-active structure in which the hybrid control was applied. The choice of this type of control stems from the fact that in the case of tracked combat vehicles in addition to the advantageous conditions of work of vehicle crew also cornering stability and the possibility of sudden acceleration or braking is important. The hybrid control allows to determine a compromise between ride comfort and stability of 2S1 platform.

Cryptographic Properties of Equivalent Ciphers

In this work the use of nonlinear filtering functions applied to Linear Feedback Shift Registers (LFSRs) to generate cryptographic sequences has been studied. Emphasis is on the class of equivalent nonlinear filter generators whose elements produce exactly the same sequence. Given a pair (filtering function, LFSR), the work develops a method of computing equivalent filtering functions applied to different LFSRs, where all these filters generate exactly the same cryptographic sequence. The security of a cipher is that of its weakest equivalent. Thus this article allows one to compute weaker equivalents of apparently secure generators.

Nonlinear filtering for map-aided navigation. Part 1. An overview of algorithms

A map-aided navigation method is considered. The main features of this method are discussed; an overview of the algorithms used to solve navigation problems is given. Considerable attention is focused on the algorithms based on the use of nonlinear filtering, which make it possible to not only formulate and solve the problem of algorithm design, but also provide a background for accuracy analysis.

Can local particle filters beat the curse of dimensionality?

The discovery of particle filtering methods has enabled the use of nonlinear filtering in a wide array of applications. Unfortunately, the approximation error of particle filters typically grows exponentially in the dimension of the underlying model. This phenomenon has rendered particle filters of limited use in complex data assimilation problems. In this paper, we argue that it is often possible, at least in principle, to develop local particle filtering algorithms whose approximation error is dimension-free. The key to such developments is the decay of correlations property, which is a spatial counterpart of the much better understood stability property of nonlinear filters. For the simplest possible algorithm of this type, our results provide under suitable assumptions an approximation error bound that is uniform both in time and in the model dimension. More broadly, our results provide a framework for the investigation of filtering problems and algorithms in high dimension.

Use of non-linear filtering for the regional–residual separation of potential field data

Abstract Regional–residual separation is essential in gravity and magnetic data interpretation and a variety of techniques have been proposed. Graphical determination of the regional allows geological information to be taken into account. Upward continuation can be used to obtain the regional field either empirically or using some hypothesis about the geology. In some cases, a matched filter can be designed and used to separate deep and shallow sources. Simple low pass filtering has also been used but without much success. Here we propose to use a non-linear filter approach to remove gravity and magnetic anomalies smaller than a given width. This technique attempts to mimic the graphical separation method. The results from synthetic models are presented as well as the results from a case study in eastern Canada and compared to regional gravity and magnetic anomalies obtained by other techniques. Contrary to the regional fields obtained by upward continuation, non-linear filtering does not have any physical meaning. However, its main advantage is that it gives a regional component of the gravity or magnetic field similar to the one obtained from a graphical separation.

On the Use of Nonlinear Filtering, Artificial Viscosity, and Artificial Heat Transfer for Strong Shock Computations

A new artificial viscosity (Q) model, based on physical conservation corrections for momentum, and a new artificial heat transfer (H) formulation are developed for the analysis of one-dimensional compressible fluid transients in plane, cylindrical, and spherical geometries. The accuracy of these formulations is verified against various benchmark shock tube problems. A Q-induced geometric error for cylindrical and spherical geometry is defined and the benefits of the Q formulation presented are demonstrated. It is also shown that these formulations can control the total variation of the solution and have superior shock-capturing capabilities. Comparisons are made with the original Q formulations of J. von Neumann and R. D. Richtmyer (1950, J. Appl. Phys.21, 232), W. F. Noh's Q&H shock-following method (1987, J. Comput. Phys.72, 78), and the piecewise-parabolic method of P. Colella and P. R. Woodward (1984, J. Comput. Phys.54, 174). The comparisons demonstrate the advantages of the new method. Numerical examples for more realistic equations of state which show the robustness of the method are also presented.

Robust motion-compensated video upconversion

The quality of field-rate conversion improves significantly with motion-compensation techniques. It becomes possible to interpolate new fields at their correct temporal position. This results in smooth motion portrayal without loss of temporal resolution. However, motion vectors are not always valid for every pixel or object in an image. Therefore, visible artifacts occur wherever such wrong vectors are used on the image. One effective method to solve this problem is the use of non-linear filtering. In this method, a wrongly interpolated pixel is either substituted or averaged with neighbouring pixels. We introduce and evaluate a new and very robust upconversion algorithm which is based on the non-linear filtering approach. It is unique in that it estimates motion vector reliability and uses this information to control the filtering process. This algorithm outperforms others in its class, especially when we have complex image sequences.

A note on the use of nonlinear filtering in computer graphics

The application of nonlinear median and alpha-trimmed mean filters to computer graphics is discussed. The filters are simple to implement, do not require large amounts of CPU time to execute, and can be applied iteratively for better control over the filtering effects. A series of examples shows that the filters eliminate spike noise in a stochastic sampling application without smearing the final image.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Application of nonlinear filtering in the real time forecasting of river flows

The practical use of nonlinear filters on conceptual stochastic watershed models is investigated. Results obtained on a case study emphasize the importance of the assumed noise components of the watershed model. Depending on such noises, runoff predictions could range from excellent to unacceptable. The simpler of the nonlinear filters considered, the extended Kaiman filter, was found to be, when effective, as good as other, more complicated filters. Although smoothing algorithms lead to improvements, their computational burden may be unacceptable.