Joint bit allocation and precoding for filterbank transceivers in NOFDM systems

Recently, the non orthogonal frequency division multiplexing (NOFDM) systems have attracted increased interest. They have several advantages over traditional OFDM systems: higher bandwidth efficiency; reduced sensitivity to carrier frequency offsets, oscillator phase noise and narrowband interference; and reduced intersymbol/intercarrier interference (ISI/ICI). In particular, low ISI/ICI will be important for future systems where Doppler frequencies will be larger (equivalently, channel variations will be faster) due to higher carrier frequencies and higher mobile velocities. In the first part of this thesis the duality of multicarrier systems and Gabor frames is discussed and applied to the design of a generalized multicarrier system based on a filterbank structure. The efficient polyphase implementation is also discussed. In this thesis the channel capacity of a GMC systems is evaluated through the diagonalization of an equivalent matrix model where intersymbol and intercarrier interferences have been included. Exploiting the majorization theory, the mutual information can be represented as a Schur-concave function and it is maximized through a joint transceiver design adding a linear precoder at the transmitter and a LMMSE equalizer at the receiver. The capacity is derived by the eigenvalue decomposition of the global system matrix including the noise colored by the receiver filtering and employing a power allocation of the transmitted power according to the well-known water-filling solution. This thesis investigates also the behaviour of the NOFDM systems when a power and bit allocation algorithm (like the Campello one) is employed in order to satisfy a certain QoS constrain. A comparison of the performances with OFDM systems is included. Finally a simple application of the cognitive radio paradigm employing filterbankbased multicarrier systems is developed and some interesting results are showed