This thesis includes two parts, the first part focuses on theoretical investigations of
als. The second part studies practical implementations of SEFDM focusing on optical,
wireless or their combination at either microwave or millimeter wave frequencies.
Chapter 2 introduces the SEFDM signal. It starts with a mathematical description
followed by a summary of existing modulation/demodulation techniques. Following
this, the self-created interference characteristics are analyzed to show challenges of
SEFDM signal recovery. Finally, the chapter includes a summary of other spectrally
efficient techniques.
In Chapter 3 efforts to simplify the signal detection are presented. Three system
architectures are proposed. The first is an iterative receiver based on the FSD detec-
tor. A new linear detection algorithm termed soft demapping iterative detection (ID)
is introduced. Following that, a hybrid ID together with FSD is presented to improve
further the performance. The second proposal is a multi-band architecture, named
block-spectrally efficient frequency division multiplexing (Block-SEFDM), for a moder-
ate size non-orthogonal system. The new architecture decomposes the whole spectrum
into several blocks. Each block can be detected separately by using an efficient de-
tector such as SD. This technique makes it practical to detect a SEFDM signal (e.g.
modulated on 128 non-orthogonal data sub-carriers). Furthermore, a detector termed
block efficient detector (BED) is presented in this work and computer simulations show
that the performance is significantly improved while the complexity is decreased by one
order of magnitude. The third proposal investigates a new iterative detection algorithm
which can simplify the detection of signals in large size non-orthogonal multicarrier sys-
tems. This work proposes a fast Fourier transform (FFT) based soft detector working
alongside a standard Bahl-Cocke-Jelinek-Raviv (BCJR) outer decoder. The detector
can iteratively improve the reliability of candidate solutions using forward error cor-
rection (FEC) based on the Turbo principle by exchanging soft information between
the FFT detector and the outer decoder. Mathematical modelling results show the
suitability of the proposed detector for use in large size (e.g. 1024 data sub-carriers)
Chapter 4 presents new applications of SEFDM in optical domain. This is the first
time to prove experimentally that SEFDM is applicable to optical communication sys-
tems. Two optical systems such as directed detection optical-SEFDM (DDO-SEFDM)
and coherent optical-SEFDM (CO-SEFDM) are designed and evaluated. The 10 Gb/s
DDO-SEFDM optical experiment can improve spectral efficiency in both electrical and
optical domains while achieving the same performance when bandwidth saving is up
to 20% or signalling rate is up to 25%. This is the first experimental verification of
the Mazo’s 25% optical faster than the Nyquist principle in the optical domain. Fur-
thermore, results indicate that 4QAM DDO-SEFDM can replace 8QAM DDO-OFDM
while achieving better performance. It is experimentally shown that a lower-order mod-
ulation format can achieve better performance by replacing a higher one. The same
results were obtained in a 24 Gb/s CO-SEFDM testbed where spectral efficiency was
further improved with extra signal processing efforts.
Chapter 5 reports the first experimental demonstration of LTE-Advanced SEFDM
signal transmission and reception over a realistic RF environment. Carrier aggregation
(CA) is a technique introduced in LTE-Advanced to achieve a higher throughput by
aggregating legacy radio resources. Meanwhile, SEFDM is a bandwidth compressed
technique that can pack more non-orthogonal sub-carriers in a given bandwidth. Con-
sidering the scarcity of radio spectrum, the SEFDM bandwidth compression technique
can be used to enhance CA performance. The combination of two techniques results in
more aggregated component carriers (CCs) in a given bandwidth. It shows that up to
7 CCs can be aggregated in a given bandwidth with guaranteed bit error rate (BER),
while OFDM can only pack 5 CCs in the same bandwidth. This chapter firstly studies
practical impairments in a realistic wireless communication environment. Multipath
fading effect is firstly investigated followed by its compensation solutions. Then, an
experimental testbed is introduced in detail. Several impairments such as imperfect
timing synchronization, phase noise and sampling clock phase offset are analyzed. Fur-
full system description and experimental setup are given together with BER results
for SEFDM and OFDM based systems using LTE-like frame and signal formats and
transmitted over an LTE standard fading channel. Experimental results firstly show
the suitability of the proposed detector for use in large size non-orthogonal multicarrier
systems. In addition, results demonstrate the bandwidth advantages of SEFDM and
confirm that the effective spectral efficiency of CA-SEFDM is much higher than that
of CA-OFDM.
Chapter 6 presented a simplified interference cancellation scheme for a non-orthogonal
multicarrier signal using a pulse shaping technique. In non-orthogonal multicarrier sig-
nals, higher spectral efficiency may be achieved at the expense of self-created ICI.
Typically, when no pulse shaping is employed, interference, contributed by all sub-
carriers, has to be minimized resulting in a receiver of significant complexity. In order
to mitigate the interference effect, the work in this chapter constrains the interference
to adjacent sub-carriers by shaping each sub-carrier with a root raised cosine (RRC)
filter, thereby suppressing the out-of-sub-carrier power leakage. Instead of cancelling
out interference from all sub-carriers, only two adjacent sub-carriers are considered in
the RRC shaped scenario. Results indicate that the newly proposed scheme can achieve
the same performance compared to the non-shaped one but with reduced receiver com-
plexity. This paves the way to non-orthogonal signal detection and non-orthogonal
CA system designs; both of importance to future wireless and wired communication
systems. A coexistence experiment for 4G and 5G signals aggregation is operated in
a realistic PCI extensions for instrumentation (PXI) environment to show the bene-
fits such as compressed bandwidth and reduced out-of-band power leakage. Practical
over-the-air testing of those proposals, targeting massive machine-type communica-
tion (mMTC), is operated on commercially developed software defined radio universal
software radio peripheral (USRP) platforms. Coexistence evaluations on two USRPs
show that SEFDM can significantly reduce interference when used with existing LTE
Nyquist-SEFDM performs well in scenarios where the spectrum is limited and in fact
it outperforms pulse shaped OFDM significantly, both in terms of bandwidth saving
and throughput.
Chapter 7 experimentally studies SEFDM performance in a 60 GHz millimeter-wave
(mm-wave) radio-over-fiber (RoF) scenario where transmission data rate is increased
without changing signal bandwidth and modulation format. Experimentally, a 2.25
Gbit/s 4QAM OFDM signal is transmitted through 250 meters of multi-mode fiber
(MMF) and then it is optically up converted to 60 GHz band at the photodiode before
delivery to a millimeter wave antenna for transmission over a 3 meter wireless link.
The work demonstrates that when the OFDM signal is replaced by an SEFDM signal
using the same modulation format and occupying the same bandwidth, the bit rate can
be increased. In addition, an 8QAM OFDM signal is experimentally evaluated due to
its equivalent spectral efficiency with 4QAM SEFDM of 33% bandwidth compression.
The purpose is to verify that a low order modulation format may replace a higher order
one and achieve performance gain.
Much work has been conceptually and practically done in terms of optimal signal
detection algorithms. To summarize, the existing detectors have trade-off issues in
performance, complexity and spectral efficiency. Some detectors show optimal perfor-
mance but with high complexity. Some detectors have low complexity but at the cost
of performance. While some detectors achieve good performance and low complexity
but with reduced achievable spectral efficiency. Therefore, the topic of efficient sig-
nal detection still remains open. Existing SEFDM detectors attempt to extract useful
information from distorted signals at the receiver, which would result in inaccurate
signal estimate since ICI has been added to each sub-carrier. In Chapter 8, alterna-
tive solutions, focusing on transmitter side, are investigated. The first solution is to
precode signals prior to the wireless channel at the transmitter, based on known ICI
information. Briefly, the technique is based on modifying the data sent on individual
to interference ratio (ScIR) of such sub-carrier as estimated from eigenvalue decompo-
sition. At the transmitter, instead of modulating data on all sub-carriers, sub-carriers
associated with high eigenvalues (good ScIR) are reserved for data and other sub-
carriers are used for precoding redundancy. At the receiver, only the sub-carriers of
high ScIR are processed for data recovery. The precoding redundancy results in ap-
parently “wasted” sub-carriers, however, the bandwidth compression characteristic of
SEFDM compensates for such “waste”. Modelling is done in simple Gaussian noise
channels and in a static frequency selective channel and for different modulation for-
mats. The second solution is straightforwardly to modulate two symbols, which have
the same absolute amplitude but opposite signs, on adjacent sub-carriers. In this case,
interference can be cancelled mutually at transmitter side. In order to further improve
performance, receiver side mutual interference cancellation could be employed without
any multiplication or matrix inverse operations. The spectral efficiency is halved by
using this method. Thus, achievable spectral efficiency has to be investigated by study-
ing various modulation formats and bandwidth compression factors. In general, an
optimal combination of modulation formats and bandwidth compression factors exists
for each achievable spectral efficiency. Practical over-the-air testing of the self interfer-
ence cancellation is operated on commercially developed software defined radio USRP
platforms. It aims at ultra-reliable and low-latency communication (URLLC) scenario
due to its low complexity and high efficiency in cancelling the self-created ICI.
Finally, Chapter 9 summarizes the work in this thesis and put forward some pro-
posals for future work.
In order to provide an intuitive understanding of the thesis structure, Fig. 1.2
depicts the background and motivation of this research work. Fig. 1.3 shows the key
work of this thesis being SEFDM with links between different research aspects.
This thesis includes three appendixes. Appendix A presents abbreviations used in
this thesis. Appendix B attached a submitted journal paper written by the author.