Research On Data Detection With Insufficient Cyclic Prefix In OFDM Systems

With rapid development of wireless communication, and more popular the wirelessapplications are, utilization of wireless spectrum are unsatisfactory. In OFDM andSC-FDMA system, the traditional method is using the cyclic prefix to eliminate theinterference, but in fact the multi-path delay is unpredictable in different environments anddifferent application scenarios. Because of the urgency of improving spectrum efficiency,OFDM systems with insufficient CP are studied, namely how to ensure the full use ofspectrum efficiency, and also eliminate the interference addressed by this insufficient CP.The traditional residual inter-symbol interference cancellation (RISIC) algorithm andcyclic prefix reconstruction (CPR) algorithm, is used as a classic method confronting ISIand ICI introduced by lack of CP, but because of its hard decision methods, the errorswould be propagated to next symbol duration. In fact many OFDM systems are codedsystems, then soft decoding method especially Turbo equalization combined withtraditional RISIC algorithms are being paid attention and much more and more attentionIn this paper a detailed analysis of ISI and ICI due to lack of CP in a variety ofscenarios, such as short-CP, no CP and full CP systems, together with fast-varying channelconditions, are discussed. And because of the uniqueness of its channel estimation (CE),we firstly propose several CE methods applied in insufficient CP systems. If the channel istoo long, designers also can do channel shortening (CS) firstly to compress the channelimpulse response to a certain length, and then, do the data detection more conveniently.CEs and CS techniques are a base for signal detection and later Turbo equalizationIn this paper an improved Turbo equalization is proposed, the initial iteration with theweighted CP reconstruction method, and frequency-domain equalization using MMSEwith SIC based FDE, the two different methods’ complexity in iterations are discussed. Inthis paper a focus on the detailed modules of the OFDM symbol detection in this kind ofTurbo equalization are also given. Simulation results show that by using the proposedmodified Turbo method, the bit error rate is reduced. Actually, if channel estimation ismore accurate and in high signal to noise ratio case, with the complex but good design ofthe receiver, the BER performance can be improved even close to that of OFDM systemwith full CP...