Viterbi Detector: Review of Fast Algorithm and Implementation

1842 days ago, 911 views
PowerPoint PPT Presentation

Presentation Transcript

Slide 1

Viterbi Detector: Review of Fast Algorithm and Implementation - Xiaohong Sheng ECE734 Project

Slide 2

Viterbi Algorithm Viterbi Algorithm: The ideal translating calculation for convolutional code, it can likewise be utilized for discourse and character acknowledgment which is demonstrated by shrouded Markov models

Slide 3

Convolutional Code Convolutional code Widely utilized for computerized correspondence Cj=[C j1 ,C j2 ] Example: G 0 C j1 G 1 Xj G 2 State C j2

Slide 4

Problems on Viterbi Algorithm Computational unpredictability increments exponentially with requirement length (state) of convolutional code Nonlinear input circle in the VA introduces a bottleneck for High speed usage Other issues, for example, Viterbi calculation is a ML (ideal) calculation if Euclidean separation is utilized. The generally utilized Hamming separation as a part of VA is sub-ideal and in this way lose some execution. On the off chance that Euclidean separation is utilized, The utilization of multiplier builds the decoder many-sided quality essentially

Slide 5

Any Solution? YES! Two primary arrangements Reduce state At slightest portion of the states can be diminished for DPSK sources. Energizing! - Yes, Believe?- ?, How? Furthermore, whatever other issue can be initiated? Pipeline Solve the bottleneck of nonlinear input? Others arrangements like Linear separation metric can be utilized Select some exceptional convolutional codes

Slide 6

Reduced state arrangement DPSK sources Received flag at the ith collector for QAM information correspondence framework can be depicted as When Xi(t) is oversampled by K, hi(t) keeps going a most extreme of d image interims and put all information from N recipients in a vector, the signs can be demonstrated as: Under the suspicions on: a)Si are orthonormal, b) Noise is Gaussian Use SVD Use Mahalanobis orthogonalization change

Slide 7

Reduced state arrangement (Cont.) It can be demonstrated that Where: So, Is influenced by info information images [St, St-1… St-k-d+1] So, the ideal location can be characterized by: It can be accomplished by VA to a M^(d+k-1) expresses, the first Rx ideal identification accomplished by VA has M^(d+k) states. Half of States is diminished

Slide 8

Pipeline Solution Pipeline M-Step Trellis (Look Ahead) or M-Step Trellis+1-Step Trellis Backward and forward Trellis

Slide 9

Other arrangements Use Linear Distances (For QPSK 8-PSK, 16-QAM) Avoid increase without losing the VA decoder execution Use doubly reciprocal convolutional codes Save 1/3 of constant operations over the VA with a state gathering and apportioning of the trellis

Slide 10

Other issues I'm considering... Could we build the decoder speed boundlessly in the event that we have limitless equipment? If not, what's most extreme speed we can accomplish? Is there ideal parcels given the extent of the source should be decoded with the goal that we can accomplish most extreme deciphering pace and utilize least equipment Woo… , Really hard numerical issue. What's more, Perhaps no arrangement Interested these issues too?

Slide 11

Reference(1) [1]. Actualizing the Viterbi calculation, Lou, H.- L. IEEE Signal Processing Magazine , Volume: 12 Issue: 5 , Sept. 1995, Page(s): 42 - 52  [2]. A decreased state Viterbi calculation for visually impaired succession estimation of DPSK sources,Tongtong Li; Zhi Ding Global Telecommunications Conference, 1999. GLOBECOM '99 , Volume: 4 , 1999 ,Page(s): 2167 - 2171 vol. [3]. A lessened state Viterbi calculation for multiuser identification in DS/CDMA frameworks ,Wang Zhaocheng; Ge Ning; Yao Yan; Qiang Wang Communication Technology Proceedings, 1996. ICCT'96., 1996 International Conference on , Volume: 2 , 1996 Page(s): 1102 - 1105 vol.2  [4]. Direct separations as branch measurements for viterbi interpreting of trellis codes,Hut-Ling Lou Acoustics, Speech, and Signal Processing, 2000. ICASSP '00. Procedures. 2000 IEEE International Conference on , Volume: 6 ,Page(s): 3267 - 3270  [5]. A limitation length based changed Viterbi calculation with versatile exertion Feldmann, C.; Harris, J.H. Correspondences, IEEE Transactions on, Volume: 47 Issue: 11 , Nov. 1999 Page(s): 1611 –1614

Slide 12

Reference(2) [6]. Many-sided quality diminishment of the Viterbi calculation utilizing doubly integral convolutional codes, Haccoun, D.; Caron, M.; Nabli, M. Correspondences, Computers and Signal Processing, 1999 IEEE Pacific Rim Conference on , 1999 Page(s): 408 –411 [7]. Elite VLSI design for the Viterbi calculation, Boo, M.; Arguello, F.; Bruguera, J.D.; Doallo, R.; Zapata, E.L. Correspondences, IEEE Transactions on , Volume: 45 Issue: 2 , Feb. 1997 Page(s): 168 - 176  [8]. Pipelined structures for the Viterbi calculation, Boo, M.; Brugera, J.D. TENCON '97. IEEE Region 10 Annual Conference. Discourse and Image Technologies for Computing and Telecommunications., Proceedings of IEEE , Volume: 1 , 1997 Page(s): 239 - 242 vol. [9]. A rapid Viterbi decoder utilizing way constrained PRML strategy and its application to 1/2 creep HD full piece rate computerized VCR, Hara, M.; Yoshinaka, T.; Sugizaki, Y.; Ohura, S. Purchaser Electronics, 2000. ICCE. 2000 Digest of Technical Papers. Page(s): 96 - 97  [10]. Novel Viterbi decoder VLSI execution and its execution, Kubota, S.; Kato, S.; Ishitani, T. Correspondences, IEEE Transactions on, Volume: 41 Issue: 8 , Aug. 1993 Page(s): 1170 –1178   [11], "A 1-Gb/s, four-state, sliding piece Viterbi decoder," P. J. Dark, T. H.- Y. Meng, IEEE J. Strong State Circuits, vol. 32, no. 6, June 1997, pp. 797-805