High Speed Turbo Decoding And Cooperative Spectrum Sensing For Next Generation Wireless Communications

Wireless communication today is developing towards higher resource efficiency,faster data transmission and processing speed, and better system reliability. However, inorder to achieve this goal, both the real-time and high-speed data processing and thescarcity and non-renewable nature of wireless spectrums, are the challenges that wemust face. Therefore, this work investigates the high-speed and low-delay Turbodecoding and the fast and reliable cooperative spectrum sensing (CSS) for nextgeneration wireless communications (B3G/4G), and obtains certain research progress tosatisfy the requirements of quality of service (QoS) for B3G/4G services. The maincontribution of this work can be summarized as follows:1. High speed Turbo decoding issues: Aiming at the defects of the conventionalhigh-speed Turbo decoding, further studies are made from two aspects, i.e., thedecoding algorithm and the decoding structure, respectively.a) Decoding algorithms are improved for high-speed and low-delay Turbodecoding. A novel radix-4SOVA algorithm is proposed and its complete mathematicalderivation is given. The key of the algorithm is to propose a novel reliability updatemethod which achieves the reliability update after combining two-step state transitionsin trellis diagram. By analyzing the range of application of the proposed algorithm, it isobtained that the algorithm is not only suitable for binary Turbo codes based on bitinterleaving, but also doubinary convolutional Turbo codes based on double-bitinterleaving. But the conventional radix-4SOVA algorithm is only suitable for binaryTurbo codes based on bit interleaving. In addition, simulation results show that the BERperformance of the novel algorithm is very close to that of radix-4Max-Log-MAP, andapproaches radix-4MAP when the extrinsic information coefficient is adaptively added.Moreover, it still has other virtues such as low decoding delay and small storageresource requirements, and achieves a good tradeoff with the computational complexity.b) Decoding structures are improved for high-speed and low-delay Turbo decoding.A novel parallel decoding scheme is proposed to decrease the decoding delay. The keyof the novel scheme is to design a novel collision-free (CF) interleaver. After theanalysis of the CF criterion, the design criterion and an implementation approach of theCF interleaver based on the window are proposed by using the idea of the slidingwindow. The novel CF interleaver allows extrinsic information immediately as eachother’s priori information to start the next component decoding so as to decreasedecoding delay. Under the condition of the same degree of parallelism and number of iterations, the comparison of time complexity shows that the decoding delay of thenovel parallel scheme is only about a quarter of that required for the conventionalparallel scheme, i.e., the novel scheme greatly decreases the decoding delay.Unfortunately, simulation results show that the novel parallel scheme suffers from theBER performance degradation compared with the conventional parallel scheme, andwith the increase in the number of dividing a subblock into windows, the performancedegrades more and more seriously. However, note that when a subblock only dividesinto two windows, the BER performance degradation of the novel parallel scheme isvery small, and its decoding delay is still about a quarter of that required for theconventional parallel scheme. Aiming at the particular case that a subblock divides intotwo windows, modified inter-subblock shuffle (M-ISbS) CF interleavers are furtherproposed, and its low-complexity architecture design for parallel Turbo decoding is alsogiven which can realize parallel outputs of the on-line interleaving addresses andsupport variable interleaving lengths. Therefore, the M-ISbS interleavers can be appliedto the novel parallel scheme where a subblock divides into two windows.2. Cooperative spectrum sensing issues: The CSS has been shown to be aneffective approach to improve the detection performance by exploiting the spatialdiversity among multiple users. Aiming at the cognitive systems of different particularscenarios, CSS schemes based on the optimal cognitive user set and the best relay areinvestigated respectively.a) Aiming at the cognitive system under correlated shadowing, the optimalselection scheme of cognitive user set is studied for CSS. First, the hypothesis testingmodel and the detection performance of CSS under the correlated log-normalshadowing scenario are introduced. Afterwards, based on whether the false-alarm andmissed-detection probabilities are constrained, three optimization problems areformulated to find the optimal set of cognitive users participating in cooperation, whichtake into account the tradeoff between detection performance and cooperation overhead.Nevertheless, it is very hard to obtain optimal solutions for them directly, known asnondeterministic polynomial (NP)-hard problems. Then the solutions using adaptivegenetic algorithms are presented, which are known as stochastic search techniquesbased on the mechanism of natural selection and natural genetics, and have successfullybeen applied to many complex optimization problems in wireless communications.Finally, a series of experiments are carried out and the results are analyzed,demonstrating that the proposed schemes can optimally select a set of cognitive users toparticipate in cooperation under the correlated shadowing scenario. b) Aiming at the cognitive system with multiple relays, the CSS schemes based onthe best relay are studied, and their detection performance over Nakagami-m fadingchannels is especially focused. First, the CSS scheme based on opportunisticamplify-and-forward relaying (denoted by CSS-OAFR scheme) is presented by usingopportunistic relay selection. The detection performance using the energy detector isderived for the CSS-OAFR scheme, and especially a tight closed-form lower bound ofthe average missed-detection probability is proposed for the convenience ofperformance evaluation in practice. By simulations, the theoretical analysis results arevalidated, and the influence of the number of relays on the detection performance is alsodiscussed. Second, the CSS scheme using the AF strategy with partial relay selection(denoted by CSS-AFS-PRS scheme) is investigated. The similar detection performanceanalysis and simulation validation are also provided for the CSS-AFS-PRS scheme. Inparticular, simulation results also show that in the CSS-AFS-PRS scheme the use ofonly two relays is enough to achieve relatively good detection performance, meaningthat the selection overhead of systems is overcome.