The reduction in both user and control plane latency is a major goal for next generation (4G) cellular networks, specifically the Long Term Evolution-Advanced (LTE-A) standard. At the same time, relay stations, which ...
详细信息
ISBN:
(纸本)9781457720529
The reduction in both user and control plane latency is a major goal for next generation (4G) cellular networks, specifically the Long Term Evolution-Advanced (LTE-A) standard. At the same time, relay stations, which introduce additional latency into transmissions, are seen as a potential means to improve cell spectral efficiency and coverage. This paper extends a number of existing relaying schemes to the case where a stricter latency constraint is applied. These low latency schemes are first evaluated for convolutional codes, for which they quickly approach the performance of full decoding at the relay with little additional latency over memoryless relaying. The schemes are then adapted for the turbo code in the LTE standard and shown to be preferable to full decoding and memoryless relaying for certain LTE-A relay types.
We propose a LDPC convolutional Code ensemble together with an expanding-window message-passing decoder that asymptotically have anytime properties when used for streaming transmission on the binary erasure channel. W...
详细信息
ISBN:
(纸本)9781457720529
We propose a LDPC convolutional Code ensemble together with an expanding-window message-passing decoder that asymptotically have anytime properties when used for streaming transmission on the binary erasure channel. We show analytically that the decoding erasure probability of these codes decays exponentially over decoding delay and determine the corresponding anytime exponents.
An efficient approach to alleviating performance loss caused by unknown (non-stationary) impulse noise, which is typically sparse and whose statistics are difficult to model accurately, was made possible in power line...
详细信息
ISBN:
(纸本)9781538663592;9781538663585
An efficient approach to alleviating performance loss caused by unknown (non-stationary) impulse noise, which is typically sparse and whose statistics are difficult to model accurately, was made possible in power line communications systems when the clipping operation is properly implemented. As reported in numerous papers, the clipping threshold, which impacts system performance profoundly, is closely related with the probability of impulse occurrence and the strength of the background noise. This paper first highlights that the least-absolute-shrinkage-and-selection-operator (LASSO) algorithm, aimed at estimating the regularization parameter λ and sparse vector of regression coefficients, can be interpreted as an effective clipping operation. Subsequently, a hybrid of the Bayesian LASSO and the maximum a posteriori algorithm using the Monte Carlo expectation maximization (MCEM) is proposed to simultaneously estimate the regularization parameter and detect the convolutional codeword in single-carrier coded systems subject to the impulse noise. Moreover, by exploiting information exchangeable between those two aforementioned entities, a novel termination rule on the EM iteration stage is devised. Numerical results attest the efficacy of the proposed scheme, despite a lack of statistical knowledge on impulse noise models.
Recently, convolutional Neural Network (CNN) based models have achieved great success in Single Image Super-Resolution (SISR). Owing to the strength of deep networks, these CNN models learn an effective nonlinear mapp...
详细信息
ISBN:
(纸本)9781538604588
Recently, convolutional Neural Network (CNN) based models have achieved great success in Single Image Super-Resolution (SISR). Owing to the strength of deep networks, these CNN models learn an effective nonlinear mapping from the low-resolution input image to the high-resolution target image, at the cost of requiring enormous parameters. This paper proposes a very deep CNN model (up to 52 convolutional layers) named Deep Recursive Residual Network (DRRN) that strives for deep yet concise networks. Specifically, residual learning is adopted, both in global and local manners, to mitigate the difficulty of training very deep networks;recursive learning is used to control the model parameters while increasing the depth. Extensive benchmark evaluation shows that DRRN significantly outperforms state of the art in SISR, while utilizing far fewer parameters. Code is available at .
In GSM communication, one of the aspects is the convolutional encoding of speech data, using a 16 state encoder. Since for every bit, additional v bits are transmitted, there is good scope for error detection and corr...
详细信息
In GSM communication, one of the aspects is the convolutional encoding of speech data, using a 16 state encoder. Since for every bit, additional v bits are transmitted, there is good scope for error detection and correction. Using state diagram at each time slot, Trellis diagrams are considered as good illustrations for the method of decoding the data and an algorithm by Viterbi is in vogue. Compared to that, the proposed technique is based on the assignment of probability values to the states at all bit timings using certain principles of fuzzy logic. By this, the length limitation of the earlier algorithm is overcome since the memory requirement is only to store the best probable fuzziness values at each time slot and hence faster and code efficient. Comparative illustrations for sample data are given. Further, the method yields the same result of the best possible data accuracy as with the trellis coding. The method has been exhaustively tested for its reliable decoding.
暂无评论