M. Moher, "Decoding via cross entropy minimization," Proceedings of the Global Telecommunications Conference, GLOBECOM '93, Houston, TX, USA, pp.809-813, December 1993.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....number of iterations needed is not known and also may vary, we need some sort of stopping criterion. Usually a fixed number of iterations is preset and we stop when the preset number is reached. This method might lead to unnecessary iterations or performance degradation if stopping too early. In [60], 61] and [62] the cross entropy between the distributions of the estimates at the outputs of the decoders at each iteration is used as a stopping criterion. This is known as the cross entropy criterion and has a low performance degradation. This idea is further considered in [63] where two ....

M. Moher, "Decoding via cross entropy minimization," Proceedings of the Global Telecommunications Conference, GLOBECOM '93, Houston, TX, USA, pp.809-813, December 1993.

....information will be used by the next MAP decoder to begin the next iteration of joint Turbo source decoding. Step 5 The above procedure repeats until some stopping criterion is met. Typically, such criteria include a prespecified maximum iteration number and the cross entropy criterion [5] [7], 10] The final output image is the processed image after the last iteration. Fig. 7 compares the error correction capability of the joint decoding system and a tandem decoding system. In this figure, the Fig. 7. Performance of EFSIF with VQ, block length = 130K, coding rate =0:5. Fig. 8. ....

....without the operation [6] Step 5 Use the modified extrinsic information for the next MAP decoder to begin the next iteration of joint Turbo source decoding. The above procedure repeats until some stopping criterion, such as fixed maximum iteration number or the cross entropy criterion [5] [7], 10] is met. Simulation results are shown in Figs. 14 and 15 in terms of error correction capability of the Turbo decoder and the quality of reconstructed images. These simulation results show that the improvement obtained by using EDSIF is comparable to (although slightly less than) that of ....

M. Moher, "Decoding via cross entropy minimization," in Proc. IEEE Globecom Conf., Houston, TX, Dec. 1993, pp. 809--813.

....(116, 464) Max Turbo Equaliser Iterations 10 No. of TDMA Frame per Packet 2 No. of Slots per TDMA Frame 1, 4 Convolutional Decoder Algorithm LOG MAP Equaliser Algorithm LOG MAP Table 1: System Parameters Different iteration termination criteria [4] such as the so called cross entropy [5] were also investigated in order to minimise the number of iteration steps for the turbo equaliser. Turbo equalisation was also proposed by Bauch and Franz [6] for the Global System of Mobile Communications known as GSM where different approaches were investigated for overcoming the ....

M. Moher, "Decoding via cross-entropy minimization," in Proceedings of the IEEE Global Telecommuncations Conference 1993, (Houston, United States), pp. 809--813, 29 November - 2 December 1993.

....systematic codes the BCJR input data can be factored into three (ideally) independent terms and that data should not been used twice in each decoding step: the concept of extrinsic information (in [18] called refinement factor ) was invented. Though this overall decoder is obviously suboptimal [19, 9, 31], a performance near the capacity bound 3 is obtained when using RSC codes, proved by extensive simulations and hardware tests [22] This empirically justifies the usefulness of iterative decoding [31] but leaves many questions unsolved, some of which are: ffl What is the structure and ....

M. Moher, "Decoding via cross-entropy minimization, " in Proc. IEEE GLOBECOM '93, Houston, Texas, pp. 809-813, Nov.-Dec. 1993.

....i.e. if the whole block was transmitted during a deep fade. Therefore, we should be able to notice whether we can improve decoding by further iterations. If not, we can stop iterating and thus reduce complexity. In the following section we will present three stop criteria. A. Cross Entropy Moher [7] has schown that the cross entropy is a useful criterion for iterative decoding. In [4] cross entropy was introduced as a stop criterion in turbo decoding. We now adopt it to turbo equalization. The cross entropy of two distributions P(x) and Q(x) is given by: log P(x) Q(x) X k ....

M. Moher, "Decoding via cross-entropy minimization, " in Globecom, pp. 809--813, IEEE, December 1993.

....A Partial factor MAP filtering algorithm is used for each component code in a given dimension, but only for the refinement factors that correspond to the filtering passes in the other dimensions. This reduces the errors introduced by the independence assumption in the MAP processing. In [56] it is shown that MAP filtering follows from entropy optimization principles. In [61] a study of iterative decoding of product codes was performed for different interleaving strategies based on combinatorial configurations. A near optimum method of iterative decoding of product codes was also ....

M. Moher, "Decoding via cross--entropy minimization," GLOBECOM 1993, Houston, Texas, USA, pp. 809--813, Dec. 1993.

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