Partial Response and Faster-than-Nyquist Signaling

Bandwidth efficiency is a key objective in the design of a digital communication system. Traditionally it is accomplished by increasing the constellation size of the signaling system. In this thesis alternative techniques are studied. Because the communication channel is continuous, so must the actual time signals be. Therefore the construction of the continuous signals is included in the code design. Bandwidth efficiency is here obtained by two different signal generation forms: partial response signaling (PRS) and faster-than-Nyquist (FTN) signaling. Common for the two techniques is that both obtain bandwidth efficiency by introducing a certain amount of intentional intersymbol interference (ISI).PRS is a coded modulation based on lowpass filtering suited for high bandwidth efficiency. The lowpass filter is designed to give optimal performance while simultaniously fulfilling bandwidth requirements. In previous work, ``optimal'' meant maximal minimum Euclidean distance. In this thesis two other objective functions than distance are investigated, bit error rate and capacity. In both cases, the obtained systems are better than those optimized for Euclidean distance.The filter introduces ISI which has to be equalized by the receiver.FTN signaling is a technique that obtains bandwidth efficiency by transmitting consecutive information carrying symbols with less time separation than required for ISI-free transmission. Thus, with FTN signaling there is no bandwidth reduction as with PRS, but rather a bit rate increase with constant bandwidth. Since the symbols appear too early, ISI cannot be avoided in the receiver. This thesis investigates the capacity limits of FTN signaling, designs concatenated coding schemes based on FTN, explores the behaviour of FTN on a MIMO channel, extends the state of the FTN art into a multicarrier setup and computes the minimum Euclidean distances for nonbinary FTN signaling.The ISI stemming from PRS and FTN signaling can be of significant duration. Optimum detection is therefore normally ruled out, and reduced complexity detection strategies have to be looked into. This thesis investigates what the decision depth of an ISI detector should be and which time-discrete ISI model should be used when a reduced complexity receiver is deployed