Baseband functions like channel estimation and symbol detection of sophisticated telecommunications systems require matrix operations, which apply highly nonlinear operations like division or square root. In this paper, a scalable low-complexity approximation method of the inverse square root is developed and applied in Cholesky and QR decompositions. Computation is derived by exploiting the binary representation of the fixedpoint numbers and by substituting the highly nonlinear inverse square root operation with a more implementation appropriate function. Low complexity is obtained since the proposed method does not use large multipliers or look-up tables (LUT). Due to the scalability, the approximation accuracy can be adjusted according to the targeted application. The method is applied also as an accelerating unit of an application-specific instruction-set processor (ASIP) and as a software routine of a conventional DSP. As a result, the method can accelerate any fixed-point system where cost-efficiency and low power consumption are of high importance, and coarse approximation of inverse square root operation is required.

In this paper, a new Minimum Description Length (MDL) approach for the characterization of a mobile phone's color camera is presented. The use of high-order polynomials, Fourier sine series, and artificial neural networks (ANN) for solving this problem are compared and contrasted. The MDL formalism is used for determining the stochastic complexity of polynomial and Fourier sine models for the characterization of a Nokia N93 mobile phone camera. A quantitative evaluation of their performances, as well as for using an ANN, is provided.

Natural scenes usually produce radiance maps that have a dynamic range much larger than the dynamic range of the imaging sensors. Due to this fact the captured images, almost always, contain under-exposed and saturated regions. Among several solutions, proposed in the open literature, the multi-frame approaches have been shown to produce high quality results by combining several shots of the same scene, captured at different exposure times. Here we introduce a low complexity multi-frame approach suitable for mobile implementations. We have implemented our method in Symbian OS in a Nokia cameraphone and the results obtained with our proposed system are shown in the paper.

Multiple input multiple output (MIMO) transmission is an emerging technique targeted at 3G long term evolution (LTE) systems. One vital baseband function in MIMO receivers is QR decomposition of the channel matrix. In this paper, a processor based complex-valued QR decomposition is presented. The processor is enhanced with complex arithmetic and inverse square root function units. The proposed processor fits well with the real-time requirements of the MIMO receiver. The computing power is tailored for typical MIMO systems. Due to the generality of the applied computing resources it can also be used for other tasks. Also, the presented principles can be applied on any customizable processor architectures to accelerate QR decomposition.

In this paper, a new Minimum Description Length (MDL) approach for the characterization of a mobile phone's color camera is presented. The use of high-order polynomials, Fourier sine series, and artificial neural networks (ANN) for solving this problem are compared and contrasted. The MDL formalism is used for determining the stochastic complexity of polynomial and Fourier sine models for the characterization of a Nokia N90 mobile phone camera. A quantitative evaluation of their performances, as well as for using an ANN, is provided.

Fixed-point simulation results are used for the performance measure of inverting matrices by Cholesky decomposition. The fixed-point Cholesky decomposition algorithm is implemented using a fixed-point reconfigurable processing element. The reconfigurable processing element provides all mathematical operations required by Cholesky decomposition. The fixed-point word length analysis is based on simulations using different condition numbers and different matrix sizes. Simulation results show that 16 bits word length gives sufficient performance for small matrices with low condition number. Larger matrices and higher condition numbers require more dynamic range for a fixedpoint implementation. Keywords—Cholesky Decomposition, Fixed-point, Matrix inversion, Reconfigurable processing.

This paper considers the implementation of multi-user detector in MC-CDMA receivers using fixed-point matrix inversion algorithms. The fixed-point word length analysis is based on the matrix condition number analysis and residual errors. The obtained bit error results have been compared to floating point matrix inversion results.

Both the matrix inversion and solving a set of linear equations can be computed with the aid of the Cholesky decomposition. In this paper, the Cholesky decomposition is mapped to the typical resources of digital signal processors (DSP) and our implementation applies a novel way of computing the fixed-point inverse square root function. The presented principles result in savings in the number of clock cycles. As a result, the Cholesky decomposition can be incorporated in applications such as 3G channel estimator where short execution time is crucial

In this paper we present an image pre-processing procedure for bar code detection in mobile devices. The goal of our method is to improve the quality of the input image, thus making bar code detection and decoding possible even in difficult situations. The implementation details and the results obtained with the proposed method on real images taken with a camera phone, are discussed

By utilizing the camera of a mobile phone as an input channel of data from printed media, there is a significant potential of interesting applications that can take advantage of this feature. We consider in this paper document image capture applications, with the purpose of text retrieval and barcode decoding. A reliable capture of document images in the mobile environment is a challenging task. Several problems, including lightening, resolution, blur and contrast, are encountered. We consider these problems, and we present a dedicated image processing chain that attempts to optimize the performance of the recognition system. We illustrate an end-to-end optimized application on Nokia mobile devices that are using this dedicated processing chain.

Fixed-point simulation results are used for the performance measure of inverting matrices by Cholesky decomposition. The fixed-point Cholesky decomposition algorithm is implemented using a fixed-point reconfigurable processing element. The reconfigurable processing element provides all mathematical operations required by Cholesky decomposition. The fixed-point word length analysis is based on simulations using different condition numbers and different matrix sizes. Simulation results show that 16 bits word length gives sufficient performance for small matrices with low condition number. Larger matrices and higher condition numbers require more dynamic range for a fixedpoint implementation. Keywords—Cholesky Decomposition, Fixed-point, Matrix inversion, Reconfigurable processing.

Fixed-point simulations for inverting matrices using transport triggered architectures are performed. Several methods are implemented in fixed-point: the Cholesky decomposition as,a direct method, Newton iterations as an iterative method, and Strassen Newton algorithm as a combined recursive method. Fixed-point implementations of these matrix inversion algorithms are tested and analyzed. A division-free implementation is targeted.

The Institute of Digital and Computer Systems is an independent research unit of Tampere University of Technology, Finland. The institute is in charge of numerous research projects, as well as both undergraduate and post-graduate teaching activities related to digital systemic and computer architectural design. The rapid growth of the number of international students, coming from both developing countries and advanced ones, with different educational background and levels of knowledge, imposes a true challenge on the teacher, especially for introductory level courses.

In this paper the problem of image restoration from its Fourier spectrum magnitude is shown to be NP-complete. We propose the use of recurrent neural networks for solving the problem. The neural network incorporates the constants related to the real and imaginary parts of the image spectrum. The solution is provided by the steady state of the neural network, then is verified and eventually improved with the iterative Fourier transform algorithm. The obtained simulation results demonstrate the high efficiency of the proposed approach.