L.Lindbom, A Wiener filtering approach to the design of tracking algorithms. PhD Thesis, Dept. of Technology, Uppsala University, Sweden, 1995.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....3. Assign an arbitrary start value as the ideal target value of V (I i ) For instance, V (I i ) # 1 if I i has a label Yes 1 if I i has a label No 4. For all training set instances I 1 , I n , update weights W 1 , W k iteratively using LMS (Least Mean Square) algorithm [WH60, Lin95] as follows: a) Initialize each W i to some small random value like 0.05. b) For each I i , i. Use the current W i to calculate V (I i ) as ii. For each weight W i , update it as W i W i # (V (I i ) V (I i ) # i where # is a small constant (e.g. 0.1) that moderates the ....

L. Lindbom. "A Wiener Filtering Approach to the Design of Tracking Algorithms With Applications in Mobile Radio Communications ". PhD thesis, Uppsala University, Nov. 1995. http://www.signal.uu.se/Publications/abstracts/a951.html. 177

....paper, in addition to the AR model, we also consider a more realistic channel model Jakes model [19] with modification of [20] We verify that a simple second order AR process (AR2) can approximate the Jakes model and can be used as a hypermodel embedded into the Kalman filter. It was observed in [21] that the spectral peak frequency of the AR2 process should be adjusted by a factor of 2 from the maximum Doppler frequency [21] This adjustment is justified in this paper. This simple AR hypermodel for the realistic fading channel reduces the complexity of the Kalman filter implementation and ....

....We verify that a simple second order AR process (AR2) can approximate the Jakes model and can be used as a hypermodel embedded into the Kalman filter. It was observed in [21] that the spectral peak frequency of the AR2 process should be adjusted by a factor of 2 from the maximum Doppler frequency [21]. This adjustment is justified in this paper. This simple AR hypermodel for the realistic fading channel reduces the complexity of the Kalman filter implementation and makes computationally tractable analysis feasible. Regarding the operation of the Kalman filter, we showed in [22] that the ....

[Article contains additional citation context not shown here]

L. Lindbom, A Wiener Filtering Approach to the Design of Tracking Algorithms with Applications in Mobile Radio Communications, Ph.D. Thesis, Uppsala University, Uppsala, Sweden, 1995. 22

....and white with zero mean. The situation is depicted in Fig. 2.1. 1 In practice, the channels will be time varying due to carrier frequency offsets and fading. The channel models, detectors and design equations presented here can easily be generalized to the time varying case, see Chapter 7 of [18]. 2 For any signal y(k) z Gamma1 y(k) y(k Gamma 1) s M (k) s 1 (k) HMN (z Gamma1 ) HM1 (z Gamma1 ) H 1N (z Gamma1 ) H 11 (z Gamma1 ) Sigma Sigma x N (k) x 1 (k) v N (k) v 1 (k) Figure 2.1: The MIMO channel model, where s j (k) is the symbol transmitted at ....

Lars Lindbom, A Wiener Filtering Approach to the Design of Tracking Algorithms, Ph.D. thesis, Uppsala University, Uppsala, Sweden, Nov. 1995.

....it is common in communication applications of signal processing to work with complex valued signals [18, 19] Such a representation carries information about both amplitude and phase. For example, several mobile radio communication filtering algorithms hinge upon complex Diophantine equations [17]. Polynomials with complex coefficients also occur in the Kharitonov s theorem, a fundamental tool for studying robust stability of linear systems [1, x6.9] Whirling shafts, vibrational systems and filters are additional examples of systems whose models involve complex coefficients [2] In the ....

L. Lindbom, A Wiener Filtering Approach to the Design of Tracking Algorithms with Applications to Mobile Radio Communications, Ph. D. Thesis, Signal Processing Group, Department of Technology, Uppsala University, Sweden, 1995.

No context found.

L.Lindbom, A Wiener filtering approach to the design of tracking algorithms. PhD Thesis, Dept. of Technology, Uppsala University, Sweden, 1995.

....the problems described above was to develop a novel class of algorithms with time invariant gains. The algorithms were required to minimize the steady state mean square parameter tracking error for smooth but fast variations in the parameters of linear regression models, in particular FIR models [8]. The proposed algorithms combine low computational complexity with an often significant performance increase as compared to LMS and windowed RLS. The increased performance is attained by introducing stochastic hypermodels which describe the second order properties of time varying parameters. ....

....; 3) where u t denotes the input data at time t. 3 The tracking algorithm The SWLMS algorithm and its design equations are presented below. The main steps for deriving the algorithm will be outlined in Section 4 and 5. Readers interested in the complete derivation are referred to the PhD thesis [8] by L. Lindbom. 3.1 The algorithm structure The basic structure of the SWLMS algorithm is given by the following equations t = y t Gamma t h tjt Gamma1 (4) h t = h tjt Gamma1 b R Gamma1 t t (5) h t 1jt = Gammap h tjt Gamma1 g 1 0 h t g 1 1 h t Gamma1 ; 6) ....

[Article contains additional citation context not shown here]

L. Lindbom, A Wiener filtering approach to the design of tracking algorithms: with application in mobile radio communications. Phd Thesis, Dep. of Technology, Uppsala University, Sweden, 1995.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC