Analysis of Parallel Sorting Algorithms in K-best Sphere-Decoder Architectures for MIMO Systems

Author(s):  
Pedro Cervantes-Lozano ◽  
Luis F. Gonz´lez-Perez ◽  
Andres D. Garcia-Garcia
2020 ◽  
Vol 11 (2) ◽  
pp. 95-102
Author(s):  
I Nyoman Aditya Yudiswara ◽  
Abba Suganda

Processor technology currently tends to increase the number of cores more than increasing the clock speed. This development is very useful and becomes an opportunity to improve the performance of sequential algorithms that are only done by one core. This paper discusses the sorting algorithm that is executed in parallel by several logical CPUs or cores using the openMP library. This algorithm is named QDM Sort which is a combination of sequential quick sort algorithm and double merge algorithm. This study uses a data parallelism approach to design parallel algorithms from sequential algorithms. The data used in this study are the data that have not been sorted and also the data that has been sorted is integer type which is stored in advance in a file. The parameter measured to determine the performance of the QDM Sort algorithm is speedup. In a condition where a large amount of data is above 4096 and the number of threads in QDM Sort is the same as the number of logical CPUs, the QDM Sort algorithm has a better speedup compared to the other parallel sorting algorithms discussed in this study. For small amounts of data it is still better to use sequential sorting algorithm.


2020 ◽  
Vol 171 ◽  
pp. 2087-2096
Author(s):  
Suneeta V. Budihal ◽  
R.M. Banakar
Keyword(s):  

Author(s):  
Mohammed Qasim Sulttan

<p>Multiple-Input Multiple-Output (MIMO) technique is a key technology to strengthen and achieve high-speed and high-throughput wireless communications. . In recent years, it was observed that frequent detecting techniques could improve the performance (e.g., symbol error rate ‘SER’) of different modern digital communication systems. But these systems faced a problem of high complexity for the practical implementation.  To solve the problem of high complexity, this work proposed Frequent Improve K-best Sphere Decoding (FIKSD) algorithm with stopping rule depending on the Manhattan metric. Manhattan metric is proposed to use with FIKSD in order to achieve the lowest complexity. FIKSD is a powerful tool to achieve a high performance close to the maximum likelihood (ML), with less complexity. The simulation results show a good reduction in computation complexity with a cost of slight performance degradation within 1dB; the proposed FIKSD requires 0% to 94% and 82% to 97% less complexity than Improved K-best Sphere Decoder (IKSD) and K-best Sphere Decoder (KSD) respectively. This makes the algorithm more suitable for implementation in wireless communication systems.</p>


Sign in / Sign up

Export Citation Format

Share Document