23rd European Signal Processing Conference (EUSIPCO)

Coding and Modulation Techniques for High

Spectral Efficiency Transmission in 5G and
Satcom
Haesik Kim

VTT Technical Research Centre of Finland,

Kaitovyl 1, FI-90571, Oulu, Finland
haesik.kim@vtt.fi

(Invited paper)

Density Parity Check (LDPC) codes [2, 3] are included. It was

Abstract Achieving high spectral efficiency is the key paradigm change. However, error correction coding schemes
requirement of 5G and Satcom systems because it provides us in 3G and 4G are almost same. Another big jump from 4G to
with much lower cost per bit. In order to achieve high spectral 5G is expected because 5G faces news challenges such as
efficiency, channel coding and modulation are the key part of the
much higher spectral efficiency and severe interferences in
physical layer. Basically, high spectral efficiency can be achieved
when adopting a high order modulation and low code rate at a heterogeneous networks. In 2008, Digital Video Broadcasting
high SNR. However, the transmit power is limited in practical - Cable 2 (DVB-C2) standard [4] was announced. It offers
wireless communication systems. The high order modulation and 16QAM to 4096QAM modulation with code rates of 1/2 to
low code rate is restrictively used. Thus, the integrated version of 9/10 for spectral efficiency of 10.8 bits/s/Hz. However, it is
5G and Satcom needs a new type of channel coding scheme. In hard to achieve under real operation conditions such spectral
this paper, we look into 5G requirements and Satcoms role in
efficiencies with acceptable effort. Basically, the transmission
5G, review candidate error correction coding schemes for 5G and
future Satcom in terms of spectral efficiency, and evaluate the rate increases will be achieved by spectral efficiency increases
performance of the candidate error correction codes. (ex. using a high order modulation). However, a high order
modulation scheme would degrade system performance
Index TermsSpectral efficiency, Error correction codes, because it is vulnerable to wireless channel impairments of a
Turbo codes, LDPC codes, 5G, Satcom, etc. practical wireless communication system. The current cellular
systems and satellite communication systems include LDPC
codes and Turbo codes. These techniques are provably
I. INTRODUCTION
suboptimal solution in terms of information theory and they

W E face a strong demand to deliver higher data rates in

various wireless communication systems, which is
driven by the growth in the video traffic and new applications.
are not optimal solution when combining with a high order
modulation technique. Thus, an alternative approach should be
investigated. It is essential to develop a new error correction
In the 5th generation (5G) systems, an integrated system of coding technique for 5G and Satellite communication
cellular networks supporting a high data rate and satellite (Satcom) systems.
communication systems covering a large area is envisioned. In Spectral efficiency represents the information rate over a
order to achieve higher data rates, we should consider many given bandwidth and can be defined as follows:
design parameters such as bandwidth, energy consumption,
latency and complexity. Sometimes they become constraint
condition. We should hurdle those barriers and achieve 5G = [bits/s/Hz] (1)
system goals.
Cellular communication systems have evolved from 1G to 5G when one coding scheme transmits R [bits/s] over an AWGN
and provided us with different services. For example, 1G of bandwidth W [Hz]. Shannon described the relationship
focused on voice services, 2G improved voice services and between the spectral efficiency and signal to noise ratio (SNR)
supported text messaging services, 3G integrated voice for the reliable transmission as follows:
services and affordable mobile internet services, 4G supports
high capacity mobile multimedia service and 5G demands < log (1 + SNR). (2)
higher capacity, lower latency and more consistent experience.
In terms of error correction coding techniques, there was a big As we can observe (2), the spectral efficiency is increased
jump from 2G to 3G because Turbo codes [1] and Low depending on SNR. Basically, high spectral efficiency can be
achieved when adopting a high order modulation and low code

978-0-9928626-3-3/15/$31.00 2015 IEEE 2796

 23rd European Signal Processing Conference (EUSIPCO)

rate at a high SNR. However, the transmit power is limited in Telecommunications Administrations (CEPT) in Europe and
practical wireless communication systems as follows: Federal Communications Commission (FCC) in US play quite
important role in the decision making process in the ITU.
SNR > 2 1. (3) They build a regional consensus even if there are national
representatives in ITU meetings. Secondly, after spectrum
A low code rate provides us with a low bit error rate (BER) allocation for a certain service, the ITU may label certain
but redundancy is increased. Thus, a high order modulation allocations for a certain technology and determine the
scheme and a low code rate can be used only when channel requirements for the technology. For the requirements some
condition is good. standardization bodies such as the 3rd Generation Partnership
In many cellular systems, the spectral efficiency is adopted in Project (3GPP) and the Institute of Electrical and Electronics
order to evaluate how efficiently the bandwidth is utilized. Engineers (IEEE) make their contributions to the ITU, where
The spectral efficiency of cellular systems (1G to 4G) has they guarantee that their technology will meet the
been increased as shown in table 1. requirements. The actual technology is specified in the internal
standardization such as 3GPP and IEEE. Some standards are
Table 1. Spectral efficiency of cellular systems. established in the regional level. The regulators are
3G responsible for spectrum licensing and protection.
1G (AMPS)
2G (CDMA200 4G The ITUs Radio communication Sector (ITU-R) Working
(GSM) 0 EV-DO (LTE)
Party 5D has the leading role and plans to deploy the 5G
Rev.A)
system in 2020 [5]. The 3GPP also plans to enhance LTE
system and meet the 5G requirements of ITU [6]. The 5G
Max link 4.08
spectral (SISO)/ target of the 3GPP is to (1) improve LTE capacity and
0.001 0.52 2.5
efficiency 16.32 (44 performance and (2) address a new business segment. The 5G
[bits/s/Hz] MIMO) timeline of ITU and 3GPP is summarized in figure 1.

As we can see in table 1, spectral efficiencies from 1G to 4G 2014 2015 2016 2017 2018 2019 2020
have been significantly improved. Especially, there was a big Report Proposal

5G Specifications
Tech Perf
jump from 1G to 2G. The role of coding and modulation Tech
ITU-R

Requrements

Workshop
Trend
techniques was significant. In addition, Multiple Input and Evaluation
Recommendation Eval criteria and
Multiple Output (MIMO) techniques are key techniques in 4G Vision method Consensus building
systems. The importance of MIMO techniques will be
emphasized in 5G systems. In order to achieve high spectral
efficiency (Beyond 5 bits/s/Hz), the integrated version of 5G Rel 13
system and Satcom needs a new type of channel coding (Public safty, Push-to-
3GPP

Rel 12 Rel 14 Rel 15 Rel 16

schemes. In this paper, we review candidate error correction talk, SDN, LTE-U,
CA, Offloading to
coding schemes for 5G and Satcom systems and evaluate their WiFi, etc.)

performance.
The remaining parts of this paper are organized as follows: In
section II, we look into 5G standardization plan and
Fig. 1. 5G timeline of ITU-R and 3GPP.
requirements of industry and academia and take a look at
Satcoms role in 5G systems. In section III, candidate channel
coding schemes for 5G and Satcom systems are introduced. In B. 5G Requirements
section IV, their performances are evaluated. Finally, section In [7], 5G system requirements are presented. The data rate
V contains the conclusion. needs to increase by roughly 1000 times from 4G to 5G. The
data traffic needs to increase by 10000 times. The 5G edge
rate (or 5% rate) ranges from 100 Mbps to as much as 1Gbps.
II. 5G SYSTEM AND SATCOMS ROLE The 5G needs to support a roundtrip latency of about 1 ms.
In this section, we discuss 5G system standardization and More than 10 to 100 times higher number of connected
requirements and then look into the role of Satcom in 5G devices are anticipated in 5G system. The energy efficiency
systems. (Joules per bit) needs to fall by at least 100 times. In addition,
accurate positioning of the devices, strong level of security,
A. 5G Standardization.
smooth mobility between different types of radio access
The 5G standardization and regulation will be developed as networks, and flexible and very dense deployments are
follows: Firstly, the international regulator such as the required. There are other views of 5G requirements from
International Telecommunication Union (ITU) defines the industry and academia as shown in table 2.
spectrum allocation for 5G services. The regional regulatory
bodies such as European Conference of Postal and

2797
 23rd European Signal Processing Conference (EUSIPCO)

Table 2. 5G requirements of industry and academia. Satellite

NTT Access Node
Samsung Academia
Nokia [9] Docomo Metis [11]
[8] [7]
[10]

Data rate 50x 10x 100x 100x 1000x

Satellite and
Cell edge Terrestrial Core
> 1Gbps 100Mbps - - > 1Gbps Network
data rate User
Terminal
Latency <1ms <1ms <1ms <5x <1ms
Terrestrial
Connected Access Node
10x 100x 100x 100x 100x
devices
10 years Fig. 2. Integrated satellite and terrestrial system.
Energy 10x longer
- M2M - <100x
efficiency battery life
battery life
As we can see figure 2, the user terminal can receive a service
Cost
10x - - - from both satellite access node and terrestrial access node
efficiency
depending on environments. For example, terrestrial access
Traffic - 10000x - >1000x node mainly covers users in urban area, the users in suburban
area receive the signal from both satellite and terrestrial access
As we can observe in table 2, academia sets more aggressive nodes, and the users in rural area (or specific area terrestrial
targets and industry sets more realistic targets. However, their access nodes cannot cover.) are covered by satellite access
gap is not so big. They approximately achieve consensus on node. When both access nodes are operated and they use the
5G requirements. same frequency bands, MIMO/Cooperative communication
Among those 5G requirements, achieving high spectral techniques can be adopted in order to improve link quality. In
efficiency is the key requirement because it provides us with [14], the key contributions of Satcom in 5G systems are
much lower cost per bit. In order to achieve high spectral described as follows: (1) coverage extension of 5G networks,
efficiency in 5G system, channel coding and modulation are (2) multimedia service improvement to the edge user, (3)
they key part of the 5G physical layer. The channel coding and terrestrial network off-loading using satellite backhaul and
modulation of the 5G system requires to achieve better traffic control via satellite and (4) network resilience by
spectral efficiency, support multi-terminal encoding and integrating satellite with terrestrial. These contributions utilize
decoding schemes in cooperative communications or the merits of Satcom systems. It will help us to meet 5G
heterogeneous networks and reduce transmission and requirements. Especially, if we allocate satellite resources to
computation power. cell edge users, we can improve cell edge data rate
improvement and optimize the infrastructure dimensioning. In
C. Satcoms role in 5G systems. addition, a high speed broadband satellite service enables us to
The 5G system targets to integration of access technologies extend reliable 5G services to high mobility users.
into one seamless experience. One important trend is
integration and interoperability of terrestrial communication III. CODING AND MODULATION FOR HIGH SPECTRAL
systems and satellite communication systems [12]. Satom EFFICIENCY
provides us with good services at specific area (sea, artic area, In this section, candidate channel coding schemes are
dessert, mountain, etc.) terrestrial communication systems do investigated for 5G and Satcom systems. Especially, a family
not cover. However, it does not compete well at terrestrial of Turbo codes and LDPC codes is focused.
communication service area and focuses on specific area it
shows off its merits such as wide range coverage, instant A. Turbo Codes and Duo-binary Turbo Codes
infrastructure and scalability. In this point of view, satellite The turbo codes achieved very low error probability which is
TV and digital broadcasting have been a big success. close to the Shannon limit. The important design criteria of the
The integration of two systems is now actively investigated. turbo codes are to find suitable component codes which
For example, Digital Video Broadcasting - Satellite services to maximize the effective free distance [15] and to optimize the
Handhelds (DVB-SH) as an integrated/hybrid satellite and weight distribution of the codewords at a low EbN0 [16].
terrestrial system is defined in [13]. Figure 2 illustrates One disadvantage of the original turbo codes is that error
conceptual network of integrated satellite and terrestrial floor occurs at BER 10-5 due to poor minimum Hamming
system. distance. possible solutions are to (1) design a good
interleaver, (2) increase the number of memories and (2) use
the non-binary turbo codes such as duo binary turbo codes or
3D turbo codes. In this section, we takes a look at duo-binary
turbo codes [17] as one candidate 5G and Satcom channel
coding. Duo-binary turbo codes are composed of two Circular

2798
 23rd European Signal Processing Conference (EUSIPCO)

Recursive Systematic Constituent (CRSC) codes with two 16QAM). However, the decoding complexity remains a major
inputs. Its information bits are encoded pair wise. CRSC obstacle for their commercial applications.
encoders do not need any tail bits.
C. Other Candidate Channel Codes
In [18], many advantages of duo-binary turbo codes are
described as follows: (1) better convergence, (2) larger In this section, we briefly take a look at two other candidate
minimum distances, (3) less sensitivity to puncturing patterns, channel codes: Sparse Regression Codes and Polar Codes.
(4) reduced latency and (5) robustness of the decoder. Sparse Regression Codes [24] as a new class of codes for
Especially, the decoder of duo-binary turbo codes can be Gaussian multi-terminal source and channel coding are
implemented much simply because the gap between MAP designed using the statistical framework of high-dimensional
algorithm and its simplified version as component decoders is linear regression. They are rate-optimal codes with low
very small. Above all things, it is very well matched with a complexity for compression and transmission. They are
high order modulation [19]. They are already adopted into suitable for cooperative communications.
several standards as shown in table 3. Polar codes [25] are first constructive coding to provably
achieve channel capacity using Kronecker product. Its
Table 3. Turbo codes and duo-binary turbo codes in standards. characteristic can be summarized as follows: (1) the block
Type of length of a polar code is normally a power of two but the code
Standard Code Rates Polynomials Termination length can be adjusted by usual code shortening methods, (2)
Turbo Codes
the rate of a polar code can be adjusted to any number
3G
(UMTS and Turbo codes
1/2, 1/3 and
13,15,17 Tail bits between 0 and 1 in increments of 1/N, (3) polar codes are
1/4 available in systematic form. (4) polar codes have a recursive
cdma2000)
structure which makes them suitable for low-complexity
4G Duo-binary
(WiMAX) Turbo codes
1/2 ~ 7/8 13, 15 Circular implementations and (5) polar codes have no error floor.

DVB-RCS
Duo-binary
1/3 ~ 6/7 13, 15 Circular IV. SIMULATION
Turbo code
In this section, the performances of duo-binary turbo codes
CCSDS Turbo codes
1/2, 1/3, 1/4
23, 33, 25, 37 Tail bits and non-binary LDPC codes are investigated. In the first
and 1/6
simulation, the frame error rate (FER) performance of duo-
binary turbo code is investigated. Simulation configuration is
as follows: 8-state duo-binary convolutional turbo codes,
B. LDPC and Non-binary LDPC
Max-log-MAP decoding, frame size 8000, 10 iterations and
In 1962, R. Gallager [20] originally invented LDPC codes in
AWGN.
his PhD thesis. However, the LDPC codes didnt get attention
until D.J.C Mackay and R.M. Neal [21] rediscovered because 0
10

the era of transistors is just started and the hardware Eta

Eta
=
=
1 (QPSK, 1/2)
2 (16QAM, 1/2)
technology does not cover the complexity of LDPC encoding -1
10 Eta = 3 (16QAM, 3/4)
Eta = 4 (64QAM,2/3 )
and decoding at that time. After the turbo codes are emerged Eta = 5 (64QAM,5/6)

in 1993, many researchers make an effort to understand how -2

10

the turbo codes achieve near Shannon limit and try to find
FER

another new error correction code. In 1996, Mackay and Neal -3

10
designed a new linear block code including many similar
features of turbo codes such as randomness, large block length -4
10
and iterative decoding. They soon realised that the new codes
are almost same as LDPC codes by Gallager. After that, -5
10
irregular LDPC codes [22] as generalization of Gallagers 0 2 4 6 8 10 12 14 16 18 20
SNR [dB]
LDPC codes are introduced by Luby in 1998. The irregular
LDPC codes became the most powerful error control codes as Fig. 3. Required SNR to achieve the spectral efficiency.
of now. When comparing with the turbo codes, LDPC codes
have several advantages. Firstly a random interleaver is not In figure 3, spectral efficiencies ( = 1, 2, 3, 4 and 5) are
required. Secondly, it has a better block error rate and a lower achieved by different modulation schemes and code rates
error floor. Thirdly, iterative decoding of LDPC codes is ((QPSK, 1/2) , (16QAM, 1/2), (16QAM, 3/4), (64QAM, 2/3),
simpler operation even if it requires more iterations. The most and (64QAM, 5/6)), respectively. Each vertical line represents
highlighted advantage is that it is patent-free. the minimum required SNR to achieve each spectral
Non-binary LDPC codes [23] can include nonzero elements efficiency. As we can observe the figure, the minimum
of H matrix like Reed-Solomon (RS) codes. Non-binary required SNR is very low. In addition, practical wireless
LDPC codes using iterative Belief Propagation (BP) show us communication systems include many other factors
good performance at medium code lengths (500 N 3000) (synchronization overhead, frequency and phase offset, etc.)
and high order modulation communications (greater than which degrade the performance. Thus, it is very challenging to

2799
 23rd European Signal Processing Conference (EUSIPCO)

meet high spectral efficiency (Beyond 5 bits/s/Hz) in practice. ACKNOWLEDGEMENT

It has been shown in [26] that the spectral efficiency of the This work was done in the 5GTN (decision number
regular (2,4) GF (2q) LDPC code increases with the field 3656/31/2014) funded by Tekes (the Finnish Funding Agency
order q, code length and modulation order. In the second for Innovation).
simulation, spectral efficiency of non-binary LDPC code is
investigated. Simulation configuration is set as follows: REFERENCES
regular (2,4) LDPC over GF(256), target BER 10-5, gray [1] C. Berrou, A. Glavieux and P. Thitimajshima, Near Shannon limit
labelling, belief propagation decoding, frame size N= 2000 error-correcting coding and decoding: Turbo-codes, Proc. Of ICC 93,
and 40000, and AWGN. Geneva, pp.1064-1070, 23-26 May 1993.
[2] R. G. Gallager, Low-Density Parity-Check Codes, Cambridge, MA: MIT
6
Press, 1963
[3] D. J. C. MacKay and R. M. Neal, Near Shannon limit performance of
5.5
low density parity check codes, Electron. Lett., vol. 32, pp. 16451646,
Aug. 1996.
5
[4] Digital Video Broadcasting (DVB); Frame structure channel coding and
modulation for a second generation digital transmission system for cable
Spectral Efficiency [bits/s/Hz]

4.5 systems (DVB-C2), DVB Document A138, June 2010

[5] http://www.itu.int/en/ITU-R/study-groups/rsg5/rwp5d/imt-
4 2020/Pages/default.aspx
[6] http://www.3gpp.org/news-events/3gpp-news/1614-sa_5g
3.5 [7] J.G.Andrews, S. Buzzi, W. Choi, S.V. Hanly, A. Lozano, A.C.K.
Soong, and J.C. Zhang, What Will 5G Be? IEEE Journal on Selected
3 Areas in Communications, vol. 32, Issue:6, pp.1065-1082, June 2014.
16QAM, N=40000 [8] Samsung Electronics DMC R&D Center, 5G Vision, Feb. 2015
256QAM, N=40000
2.5 16QAM, N=2000
[9] Nokia, 5G use cases and requirements, networks.nokia.com
256QAM, N=2000 [10] NTT Docomo, 5G: Next-Generation Mobile Communications System,
2 2014
3 4 5 6 7 8 9 10 11 12 13
Eb/N0 [dB]
[11] Metis Deliverable D1.1, Scenarios, requirements and KPIs for 5G
mobile and wireless system, Document Number: ICT-317669-
Fig. 4. Non-binary (2,4) LDPC codes over GF(256). METIS/D1.1.
[12] B. Evans, M. Werner, E. Lutz, M. Bousquet, G.E. Corazza, G. Maral, R.
As we can observe in figure 4, spectral efficiency increases Rumeau and E. Ferro, Integration of satellite and terrestrial systems in
future multimedia communications, IEEE Wireless Communications,
according to modulation order and frame size. Non-binary vol. 12, issue:5, Oct. 2005.
LDPC code with the large codeword lengths achieved high [13] Annex 12 to ITU-R Working Party 4B Chairmans Report on the
spectral efficiency (Beyond 5 bits/s/Hz) at relatively low Twenty Sixth Meeting. Document 4B/85, 27 May 2009.
[14] NetWorld2020s SatCom WG The role of satellites in 5G, July 2014.
power. This confirms that results obtained in [26]. [15] D. Divslar and R.J. McEliece, Effective Freee Distance of Turbo
From two simulations, we realized duo-binary turbo codes Codes, Electronic Letters, vol. 32, no. 5, pp. 445-446, 29th Feb 1996.
[16] D. Divsalar and F. Pollara, One the Design of Turbo Codes, TDA
and non-binary turbo codes can achieve high spectral Progress Report 42-123, Jet Propulsion Lab, Pasadena, California,
efficiency at relatively low power. However, the required pp.99-121, 15th November 1995.
transmit power of 5G is much lower than we expect. Thus, [17] C. Douillard and C. Berrou, Turbo Codes with rate-m/(m+1)
constituent convolutional codes, IEEE Transactions On
further improvements of both error correction coding schemes Communications, vol. 53, pp.16301638, Oct 2005.
are needed. [18] C. Berrou, M. Jzquel, C. Douillard, S. Keroudan and L. Conde
Canencia, Duo-binary turbo codes associated with high-order
modulations, 2nd ESA Workshop on Tracking Telemetry and Command
V. CONCLUSION Systems for Space Applications, TTC2001, Noordwijk, the Netherlands,
Oct. 2001.
High spectral efficiency (Beyond 5 bits/s/Hz) is one of key
[19] S.L. Goff, A.Glavieux and C. Berrou, Turbo-codes and high spectral
aspects in 5G and Satcom systems. When we design 5G and efficiency modulation, Proc. of IEEE International Conference on
Satcom systems, there are several constraints such as Communications (ICC) 1994, pp. 645-649, May1994.
bandwidth, energy consumption, latency and complexity. We [20] R. G. Gallager, Low Density Parity Check Codes, Monograph, M.I.T.
Press, 1963.
should hurdle those barriers and achieve the goal. Error [21] D.J.C. MacKay and R. M. Neal, "Near Shannon Limit Performance of
correction codes as one key technique of 5G and Satcom will Low Density Parity Check Codes," Electronics Letters, July 1996.
be able to improve spectral efficiency of 5G and Satcom [22] M.G. Luby, M. Mitzenmacher, M.A.Shokrollahi, and D.A. Speilman,
Improved low-density parity check codes using irregular graphs and
systems. Error correction codes of 5G systems requires to belief propagation, IEEE Int. Symp. on Information Theory (ISIT), 16-
achieve a better spectral efficiency, support multi-terminal 21 Aug, 1998.
encoding and decoding schemes in heterogeneous networks, [23] M. C. Davey and D. Mackay, Low-density parity check codes over
GF(q), IEEE Comm. Letters, vol. 2, pp. 165167, June 1998.
and reduce transmission and computation power. The [24] A. Barron and A. Joseph, Toward fast reliable communication at rates
integrated system of 5G and Satcom should find a new type of near capacity with Gaussian noise, Proc. of IEEE ISIT 2010.
channel coding scheme. In this paper, we reviewed several [25] E. Arikan, Channel polarization: A method for constructing capacity
achieving codes for symmetric binary-input memoryless channels,
candidate error correction codes for 5G and Satcom systems IEEE Trans. Inf. Theory, vol. 55, pp. 3051 3073, July 2009.
and evaluated their performances in terms of spectral [26] D. Declercq, M. Cloas and G. Gelle, Regular GF(2 q)-LDPC
efficiency. modulations for higher order QAM-AWGN channels, Proc. Of
International Symposium on Information Theory and its Applications
(ISITA) 2004, Parma, Italy, 10-13 Oct. 2004.

2800