Study On The Key Technologies Of OFDM And Cognitive OFDM Systems

With the advent of new wireless applications with high data rate, as well as the growth of existing wireless services, demand for additional bandwidth is rapidly increasing and the scarcity of spectrum is more and more serious. However,actual measurements of Federal Communication Commission(FCC)indicate that the truth of spectrum scarcity lies in the inefficient usage of the licensed spectrum. Cognitive radio(CR) technologies have been proposed for solving the contradiction between spectrum scarcity and spectrum underutilization. In Cognitive radio networks(CRNs), CR users detect the presence or absence of the licensed users(LUs) periodically and access vacant spectrum opportunistically without causing excessive interference to the LUs. The utilization of radio spectrum resource can be improved significantly by employing the dynamic resource allocation.As a multi-carrier technique with high bandwidth efficiency, capability of combating multi-path, orthogonal frequency division multiplexing(OFDM) has been widely used in broadband wireless communication. OFDM offers a great flexibility in spectrum shaping by simply turning on and off individual subcarriers and achieves high data rate via collective usage of a large number of non-contiguous subcarriers, which can efficiently exploit the idle spectrum of LUs. The special OFDM whose frequency band is not contiguous is termed as non-contiguous OFDM(NC-OFDM), which has been the most promising candidate for the Physical Layer(PHY) technology of CR networks. However, since OFDM was first proposed, its inherent problem of high peak-to-average-power ratio(PAPR) has not been resolved. The better performance of dynamic resource allocation algorithm considering the fairness of users sparks us to further study. The proposition of cognitive OFDM system is of great significance for solving the problem of spectrum scarcity. However, in order to realize the coexistence of cognitive system and licensed system, the out-of-band radiation must be suppressed. At the same time, the study of dynamic resource allocation in cognitive OFDM system plays an important role in improving spectrum efficiency. Aiming at these problems, an in-depth investigation of non-cognitive OFDM and cognitive OFDM system is implemented in this dissertation from the perspectives of PAPR reduction, sidelobe suppression, and dynamic resource allocation, respectively. The primary contributions included are summarized below:1. To reduce the inherent high peak-to-average ratio(PAPR) problem of OFDM signals, a novel threshold-based piecewise companding transform is proposed with moderate BER performance loss in the dissertation. Based on the statistical characteristic of amplitudes, OFDM signals are classified into three groups: small(amplitude) signals, average(amplitude) signals and large(amplitude) signals. Two suitable thresholds are dedicatedly designed, where one is to compress the large signals for reducing peak values, and the other is to amplify the small signals for raising the average power levels, thus reducing the PAPR of OFDM signals. At the receiver, companding noise is estimated and cancelled by introducing the iterative detection, which would, in turn, result in the decrease of the in-band distortion caused by the piecewise companding transform. Accordingly, a lower bit error rate(BER) performance loss can be obtained. As a special form of OFDM, NC-OFDM also suffers from high PAPR. However, due to the noncontiguous nature of the spectrum in CR systems, the statistical properties of the PAPR of NC-OFDM signals are different from that of OFDM signals. They are investigated in the dissertation, which is expected to help to design the PAPR reduction algorithms for NC-OFDM signals.2. Adaptive priority-based subcarrier allocation algorithms for multiuser non-cognitive OFDM systems are investigated. Firstly, the effect of users’ priority in selecting subcarriers on the system capacity is studied with two users as an example. On the basis, conflict-resolution subcarrier allocation algorithms are proposed. Initialization in the proposed algorithm is independently done by searching higher channel gain subcarriers for each user. Conflict will happen when two or more users choose the same subcarrier. How to resolve conflicts? Average channel gain is used as an index to determine which user has higher priority. Two different priority rules are defined where the larger the average channel gain is, the higher priority is, or the smaller the average channel gain is, the higher priority is. Accordingly, two conflict-resolution algorithms are presented. Simulation results show that the proposed method can approximately achieve the upper bound of system capacity when the users with lower average channel gain have higher priority.3. To suppress the out-of-band(OOB) radiation of OFDM signals, a constellationexpansion-based sidelobe suppression scheme for cognitive OFDM systems is proposed. Since each original symbol is independently expanded in the conventional constellation expansion(CE) method, it is difficult to control the adjacent expanded constellation points. Once adjacent symbols are expanded to the constellation points which have the same sign over the real or imaginary component, it will have negatively impact on the sidelobe power reduction. Different from CE, the proposed scheme is based on symbol pair, rather than a single symbol. Two adjacent symbols in the original sequence are firstly grouped into a pair. The symbols in each pair are then expanded to new constellation points which are out of phase by180 or nearly by 180 for reducing sidelobes. At the receiver, the mapping table at the transmitter can be taken full advantage of to correct decision error, which will reduce the BER performance loss due to the use of higher order constellation set. Theoretical analysis and simulation results show the effectiveness of the scheme.4. Resources in CRNs should dynamically be allocated according to the sensed radio environment. Firstly, the interference to LUs is carefully analyzed caused by the CR network due to imperfect spectrum sensing and OOB emissions from CR subcarrier sidelobes. Since the activities of LUs play an important role in evaluating resource allocation algorithm, two kinds of traffic pattern of LU occupying channels are then introduced, which are deterministic traffic and stochastic traffic. On the basis, the problem of resource allocation in CRNs can be formulated. Two heterogeneous services exist in the CRNs which are real-time service(or delay sensitive service) and non-real time service(or delay tolerant service). The CR users are accordingly divided into delay tolerant-cognitive users(DT-CUs) and delay sensitive-cognitive users(DS-CUs). Since DT-CUs are concerned with the throughput, while the DS-CUs are more concerned with the switching delay, the resource allocation problem is formulated as a bi-level programming problem(BLPP). We aims to minimize the switching delay of DS-CUs with minimum-rate guarantee at the higher level, and we try to maximize sum capacity of DT-CUs at the lower level. Considering the computational complexity of solving the problem of BLPP, we tackle this problem into two steps with efficient heuristic algorithm for sub-channel assignment and power enhancement. Simulation results are presented to demonstrate that our proposed resource schemes work quite well forthe concerned scenarios.