Research Paper Journal of Engineering and Technological Sciences An Enhanced Dynamic Spectrum Allocation Method on Throughput Maximization in Urban 5G FBMC Heterogeneous Network Nurzati Iwani Othman1,*.
Ahmad Fadzil Ismail1.
Khairayu Badron1.
Wahidah Hashim2.
Mohammad Kamrul Hasan3 & Sofia Pinardi4
\\\\\\\\
Department of Electrical & Computer Engineering.
International Islamic University Malaysia.
Jalan Gombak, 53100.
Selangor.
Malaysia College of Computer Science & Info Tech.
Universiti Tenaga Nasional.
Kajang.
Selangor.
Malaysia Faculty of Information Science & Technology.
Universiti Kebangsaan Malaysia, 43600 UKM Bangi.
Selangor.
Malaysia Electrical Department.
Universitas Muhammadiyah Prof.
Dr.
Hamka.
Jalan Tanah Merdeka No.
Kp Rambutan Jakarta 13830.
Indonesia Corresponding author: nurzati.
iwani90@yahoo.
Abstract Reports have shown that the demand for data managed by wireless systems is expected to grow by more than 500 exabytes by 2025 and beyond.
5G networks are predicted to meet these demands, provided that the spectrum resources are well In this paper, an enhanced dynamic spectrum allocation (E-DSA) method is proposed, which incorporates a cooperative type of game theory called the Nash bargaining solution (NBS).
It was assumed that there is one primary user (PU) and two secondary users (SU) in the network and their spectrum allocation was analyzed by testing the validity of the algorithm itself by using price weight factors to control the costs of the spectrum sharing.
The solution was established by combining a proposed multiplexing method called the Filter Bank Multicarrier (FBMC) for 5G configuration, with the E-DSA algorithm to maximize the throughput of a heterogeneous 5G network.
It was shown that the throughputs for 5G with EDSA implementation were always higher than those of the ones without E-DSA.
The simulation was done using the LabVIEW communication software and was analyzed based on a 5G urban macro and micro network configuration to validate the heterogeneity of the network.
Keywords: Enhanced Dynamic Spectrum Allocation (E-DSA).
Filter Bank Multicarrier (FBMC).
heterogeneous network.
Offset Quadrature Amplitude Modulation (OQAM).
Nash Bargaining Solution (NBS).
Introduction Background Study A 5G network configuration incorporates several different systems, i.
Heterogeneous Network (H-Ne.
Internet of Things, relay node, millimeter-wave and device-to-device communication, as mentioned by Siddiqui et al.
, which creates a multi-tier H-Net, as stated by Pirinen .
Figure 1 shows a representation of the 5G HNet Network.
According to Haroon et al.
, each tier requires the least amount of energy and power for transmission.
If intertier and intra-tier interferences are adequately handled, the use of several tiers in a cellular network architecture will result in superior performance in terms of capacity, coverage, spectral efficiency, and overall power consumption, as explained by Hossain et al.
and Fooladivanda and Rosenberg .
A study conducted by Han et al.
reported that different transmission powers are frequently used by base stations at different tiers.
Thus, it is necessary for the 5G network to provide the users quality of service, especially when they are travelling at high speeds, as mentioned by Bogale and Le .
Copyright A2023 Published by IRCS - ITB
ISSN: 2337-5779
Eng.
Technol.
Sci.
Vol.
No.
1, 2023, 40-51
DOI: 10.
5614/j.
Dynamic Spectrum Allocation Method DOI: 10.
5614/j.
In a dense network configuration, in order for the user to be able to receive the extracted data from the multiplexed signal, a multiplexing method is required.
Faizan et al.
mention that this method is a process of merging multiple signals into a single signal.
The Orthogonal Frequency Division Multiplexing (OFDM) is the multiplexing method that is currently being implemented for 4G.
Figure 2 shows OFDMAos block diagram.
According to Prasad et al.
OFDM eliminates Inter-symbol Interference (ISI) by distributing the overall bandwidth into orthogonal subchannels that transforms frequency-selective multipath fading into flat fading.
Figure 1 Representation of 5G H-Net.
Figure 2 OFDM block diagram.
A promising approach to meet future bandwidth requirements, mentioned by Kumar and Payal .
, is also important to achieve seamless communication.
For 5G.
Khudair and Singh .
considered FBMC to be the best choice as multiplexing method among other 5G waveform contenders such as Filtered-OFDM (F-OFDM), studied by Nekovee et al.
Universal Filtered Multicarrier (UFMC), studied by Roessler .
and Generalized Frequency Division Multiplexing (GFDM), studied by Anish et al.
, mainly because of FBMCAos low out-of-band (OOB) radiation, as mentioned by Baltar et al.
According to Sangdeh and Zeng .
OFDM systems suffer from high out-of-band radiation originated from the sidelobes of the modulated subcarriers, unlike in FBMC, as explained by Kumar and Bharti .
The FBMC configuration is shown in Figure 3.
Nurzati Iwani Othman et al.
Figure 3 FBMC block diagram.
The transceiver block diagram of the typical FBMC multiplexing method is illustrated in Figure 3 .
dapted from Franzin and Lopes .
The key modification is the substitution of OFDM with a multi-carrier system with the implementation of filter banks in the polyphase network section.
The prototype filters (PF) in the filter banks need to be carefully designed to obtain a more enhanced spectral shaping of subcarriers as compared to OFDM.
In addition to that, the prototype filter also promises an efficient spectral utilization by lessening interference among subcarriers.
The transmission bandwidthAos maximum capacity can also be achieved in FBMC configuration by implementing Offset Quadrature Amplitude Modulation (OQAM).
Related Works Based on the review by Azari and Masoudi .
, 5G H-Nets can have issues with several interferences that cause attenuation, fading, reflection, refraction, scattering and shadowing effects.
As a result, implementing an effective interference mitigation mechanism for 5G is critical.
Qamar et al.
have already started investigating several interference management approaches for 5G.
One of the approaches is the hybrid inter-cell interference coordination, studied by Huang et al.
Another comprehensive study, by Alzubaidi et al.
, determined that modeling methods for interference management are also effective.
Such methods are Poisson cluster process, proposed by Yang et al.
, and stochastic geometry, proposed by Tarriba-Lezama and ValdezCervantes .
However, these methods produce a large processing burden and lack appropriate backhaul.
Thus, in this study, the Dynamic Spectrum Allocation (DSA) technique was found to be the best interference mitigation technique for 5G.
Previous researchers have proposed several dynamic spectrum allocation strategies for 4G, such as Hasan et al.
and Othman et al.
, to improve the effectiveness of heterogeneous networks in 4G.
Then.
Rony et al.
improved the technique for 5G to promote proper utilization and fair distribution, following the dynamic traffic requirements of each cell.
However, the study did not thoroughly focus on the setup of the heterogeneous To substantiate future data requirements and support a diverse set of devices, 5G networks are predicted to meet user demands with competently managed spectrum resources.
Hence, an enhanced spectrum allocation technique (E-DSA) is an essential tool for optimizing the spectrum management practices for 5G.
Several DSA techniques were studied and compared in order to choose the best type of DSA for 5G.
Table 1 shows a summary of existing 5G DSA techniques.
Table 1 Summary of 5G DSA techniques.
Author and year Dong.
Method Koley.
, 2019 .
Markov models Jaishanthi.
, 2019 .
Gayathri.
, 2016 .
Multi agent system (MAS) Game theory (GT) Auction Strengths Well adapted to competitive Well adapted to modeling and prediction of the channel behavior Well adapted to modeling the interaction between users Well adapted to both competitive and cooperative environments Limitations Not well adapted to cooperative Not well adapted to modeling the interaction between users Must implement other techniques to perform more complex processing Does not consider how players should interact to reach this equilibrium Dynamic Spectrum Allocation Method DOI: 10.
5614/j.
In this study.
GT with NBS properties was chosen as the E-DSA method because it is a cooperative game that optimizes the objective functionsAo product instead of computing its total, which may result in a larger pay-off per user, as mentioned by Han et al.
The rest of this paper is organized as follows: Section 2 discusses the system configuration and parameters for the FBMC and E-DSA configurations as well as the mathematical expressions used for BER and throughput Section 3 presents the results and discussions and, lastly.
Section 4 provides the conclusion.
System Configuration and Parameters Proposed FBMC Configuration for 5G The assumed parameters for FBMC configuration using Labview Communications are summarized in Table 2.
Table 2 Assumed parameters for FBMC using Labview Communications software.
Parameters Prototype Filter IQ rate (H.
Bandwidth (H.
Subcarrier spacing QAM order FFT size Number of subcarriers Channel Estimation Description Lowpass Windowed FIR Linear mean square The transmitter configuration for FBMC is like that of OFDM, except additional programming blocks are added into the configuration, namely OQAM pre-processing and Synthesis Filter Bank blocks, and there is no Cyclic Prefix.
The simulation design for FBMC Transmitter is shown in Figure 4 below.
Figure 4 FBMC transmitter configuration.
The receiver configuration for FBMC, shown in Figure 5, is similar to that of OFDM, except additional programming blocks are added into the configuration, namely the Analysis Filter Bank and OQAM postprocessing blocks.
Nurzati Iwani Othman et al.
Figure 5 FBMC receiver configuration.
Proposed E-DSA Design and Parameters The system model for the present research is shown in Figure 6.
It is assumed that one primary user (PU) system and N SUs were incorporated into the spectrum distribution system.
The SUsAo base stations and the total spectrum of the PU system are denoted by Bfull.
To obtain additional revenue, the PU system auctions free spectrum resources to the i-th SU at unit bandwidth price, which is a function of the spectrum price.
It is worth noting that the size of spectrum Bempty (Bempty O Bful.
provided by the PU for SUs varies from time to time.
Figure 6 System model for E-DSA 5G FBMC network.
Dynamic Spectrum Allocation Method DOI: 10.
5614/j.
Firstly, the SUs obtain the available spectrum information of the PU through spectrum sensing, including the spectrum quality and the available spectrum size, and transmit it to the base station.
Then, under the guidance of the base station, the SUs obtain the spectrum through the Nash bargaining method to meet the communication requirements, and maximize the total revenue of the SU system.
According to Han et al.
the description of spectrum allocation using the Nash bargaining scheme are explained in Table 3 below:
Table 3 Description of spectrum allocation using the Nash bargaining scheme.
Item Description Set of secondary users (SU.
Set of allocations where SUs cooperate with each Ui,min Minimum revenue that the i-th SU is required to gain
(A,
Umi.
N-person bargaining game problem Remarks A1, 2,.
Revenue received by the i-th SU < A No cooperation for i-th SU Ui A 0.
(U1,min,U2,min,.
,UN,mi.
{U Ea A | U C Umin ,AiEaN} is a non-empty bounded In this paper, the following optimal solution is used based on .
ycOycn OO,ycOycn OuycOycn,ycoycnycu ,OAycn OcycA ycn=1.
cOycn Oe ycOycn,ycoycnycu ) .
When the revenue of each SU satisfies ycOycn Ou ycOycn,ycoycnycu , the SUs will cooperate with each other.
Therefore, it is assumed that ycOycn,ycoycnycu = 0, i.
, ycOycn > 0.
To obtain spectrum from the primary user (PU) system, the SUs are required to pay a cost to the PU by means of bargaining.
Thus, this study contains two sections to define the utility function, firstly, the revenue ycUycn .
caycn ) = OIycn yuCycn ycaycn determined once the i-th SU has been allocated to spectrum bi and secondly, the payment of cost ycsycn .
caycn ) = ycyycn ycaycn that the i-th SU is responsible for.
The utility function is defined as ycOycn .
caycn ) = ycUycn (OIycn , ycaycn ) Oe ycs.
cyycn ycaycn ) = OIycn yuCycn ycaycn Oe ycyycn ycaycn where yuCycn = log 2 .
yayuycn ) is the spectrum efficiency function of the i-th SU, ya = .
yuA , and OIycn is the revenue factor of unit transmission rate of the i-th SU.
OIycn is inversely proportional to the bandwidth request size, which results in OIycn = ycu yc( ), where x and y are constants.
Furthermore, yuCycn ycaycn is the throughput, yuA is the target BER, ycaycn and yuycn is the SINR of the i-th SU.
Therefore, the SUsAo utility function is formulated as:
ycOycn .
caycn ) = ycaycn yuCycn .
cu yc ( )) Oe ycaycn ycaycn (OcycA ycn=1 ycaycn ) ycaycn Eq.
can be optimized by solving the model under the following constraints:
ca1 ,yca2 ,A,ycaycA ) ycO = OcycA ycn=1 ycOycn ycaycn yuCycn Ou ycycn,ycoycnycu OcycA ycn=1 ycaycn O yaAyceycoycyycyc OAycn = 1,2, .
, ycA where ycaycn is the bandwidth size obtained in the solution, yuCycn is the spectrum efficiency, and ycaycn yuCycn is the throughput The constrained optimization problem of Eq.
can be solved by using the Lagrange multiplier extremum method according to Kuhn-Tucke theory.
The Lagrange function M is formulated in Eq.
ycA = OcycA ycn=1.
caycn yuCycn .
cu ) Oe ycaycn ycaycn (Ocycn=1 ycaycn )) Oe yuN(Ocycn=1 ycaycn Oe yaAycnyccycoyce ) Ocycn=1 yusycn .
caycn yuCycn Oe ycycn,ycoycnycu ) ycaycn The E-DSA solution for the i-th SU can be solved via a series of the Lagrange multipliersAo iterations.
Eq.
defines the solution of the spectrum request strategy ycaycn , .
cN .
cN) ycA cN) yuC yus ycuyuCycn Oeycaycn (OcycA ycn ycn ycOycn ycayc )OeOcycOycn ycayc ycayc OeyuN Nurzati Iwani Othman et al.
The assumed parameters of the E-DSA algorithm are listed in Table 4.
Table 4 Assumed parameters for E-DSA algorithm.
Parameters Total spectrum of PU.
Bfull
x, y
Minimum rate requirement, ycycn,ycoycnycu Initial value of Lagrange multipliers Price weight factors, c1=c2 Description 20 MHz, 0 C Bempty C Bfull
5, 1
2 Mbps.
yuN .
= 10, yusycn = 5.
1, 2
Proposed Theoretical Bit Error Rate (T-BER) and Throughput Calculations for 5G Urban Macro and Micro Configurations Several analysis methods were reviewed on how to analyze the proposed configurations, such as the spectral density analysis, as performed by Handyarso and A.
Kadir .
, as well as the study conducted by Susanti et al.
, and throughput analysis, as studied by Hmamou et al.
In this study, a throughput analysis was chosen.
The set of formula was implemented from the 3GPP technical report by ETSI .
to calculate the throughput of 5G FBMC configuration with and without E-DSA implementation.
Table 5 lists the assumptions for the parameters for the throughput calculation.
Table 5 Assumed parameters for throughput calculation.
Parameter Distance between TX (Macrocell BS) and RX (Microcell UE) dma,mi.
Distance between TX (Microcell BS) and RX (Macrocell UE) dmi,ma.
Values 15 Ae 0.
01 Ae 0.
The path loss between macrocell users and the macrocell base station denoted by ycEyaycoyca , is formulated in Eq.
ycEyaycoyca = 28.
0 22 log10 .
ccycoyca ) 20 log10 .
ceyca ) .
where yccycoyca is the distance between transmitter and receiver for the macrocell network.
Besides that, the path loss in the microcell network, denoted by ycEyaycoycn , is formulated as follows:
ycEyaycoycn = 32.
4 21 log10 .
ccycoycn ) 20 log10 .
ceyca ) .
Secondly, the channel gain for both the macrocell and the microcell network is formulated as follows:
OeycEya ya = 10 10 After that, the signal to interference plus noise ratio (SINR), yuyeayeC for the macrocell network is as follows:
yuycoyca = yaycoyca,ycoyca yycEycoyca yua 2 (Ocycuyceycnyci.
yaycoyca,ycoyca yycEycoyca ) .
aycoyca,ycoycn yycEycoycn ) .
where ycEycoyca is the transmit power of the microcell base station, yaycoyca,ycoyca is the channel gain between a microcell user and the macrocell base station and yua 2 is the power of the AWGN.
Furthermore, the yuycoycn for the microcell network when considering the interference caused by neighboring cells as follows:
yuycoycn = yaycoycn,ycoycn yycEycoycn yua 2 (Ocycuyceycnyci.
yaycoycn,ycoycn yycEycoycn ) .
aycoycn,ycoyca yycEycoyca ) .
Then, before calculating the throughput, the total capacity of users is formulated as follows:
ya = OIyce y log 2 .
where OIyce is the subcarrier spacing and yu = -1.
5*ln.
relates to the bit error rate (BER).
Au.
The T-BER expression used for OFDM modulation in this paper was taken from Nissel and Rupp .
For the same bandwidth.
Mestoui Dynamic Spectrum Allocation Method DOI: 10.
5614/j.
and El Ghzaoui .
implied that the SINR of OFDM is yuycCyayaycA = .
uyayaAycAya ), because FBMC only experiences half the noise power.
Thus, the BER expression of FBMC becomes the following expression:
yuAyayaAycAya = .
Oo1 yu Finally, for a typical 5G FBMC configuration without E-DSA implementation, the throughput of a serving macrocell is formulated as follows, ycNycEycycycy = Oc.
u y y.
where is the subcarrier assignment and is set to 1.
Result Analysis Validity of E-DSA: Effect of SUsAo Price Weight Factors on Spectrum Allocation Figure 7 depicts the effect of SUsAo price weight factors on spectrum allocation for the first and second SUs for both S-BER and T-BER bit error rates.
The bandwidths obtained by the first SUs for all conditions are always higher than those of the second SUs, that is, when the price factor is set to 1 for the first SU and 2 for the second SU.
This is because the higher the price weight factor, the higher the unit spectrum price, which may lead to overpayment of costs for the same bandwidth.
For these conditions, their own bandwidths will be reduced Thus, when there is a sudden increment in the SUAos bandwidth request, the SUAos price weight factor is increased by the PU to prevent from the SU gaining an excessive spectrum.
Besides that, the S-BER with E-DSA implementation has the highest spectrum requests followed by the T-BER with E-DSA implementation and lastly the T-BER without E-DSA implementation.
THOUSANDS
Spectrum Request of 1st SU (MH.
- S-BER (WITH E-DSA) Spectrum Request of 1st SU (MH.
- T-BER (WITH E-DSA) Spectrum Request of 1st SU (MH.
- T-BER (WITHOUT E-DSA) Spectrum Request of 2nd SU (MH.
- S-BER (WITH E-DSA) Spectrum Request of 2nd SU (MH.
- T-BER (WITH E-DSA) Spectrum Request of 2nd SU (MH.
- T-BER (WITHOUT E-DSA) SPECTRUM REQUESTS (HZ) 12 13 14
ITERATION
Figure 7 Spectrum requests from the first and the second SU.
Nurzati Iwani Othman et al.
Throughput Analysis Figure 8 shows the throughputs calculated from the S-BERs and T-BERs for 5G FBMC configurations with and without E-DSA implementation.
By comparing the average throughput increments between the typical 5G FBMC and E-DSA 5G FBMC configurations, it is proven that the E-DSA for 5G FBMC produces higher improvements than that of the typical 5G FBMC configuration, i.
, by 114%.
When implementing the simulated BER values, the throughput was found to be 4% higher than that when using the theoretical BER values.
This shows that the throughput of the 5G configurations can be improved by combining the proposed 5G FBMC design using the software with the proposed E-DSA algorithm.
Table 6 summarizes the average percentage of throughput enhancement for the configurations.
X 10000
TP for T-BER .
G FBMC - NO eDSA) TP for S-BER .
G FBMC - NO eDSA) TP for T-BER .
G FBMC - WITH eDSA) TP for S-BER .
G FBMC - WITH eDSA) THROUGHPUT
SINR
Figure 8 Throughput analysis for T-BER and S-BER 5G FBMC configuration with and without E-DSA Table 6 Average percentage of throughput enhancement when implementing E-DSA for 5G FBMC.
Item
Theoretical BER
(T-BER)
Simulated BER
(S-BER)
Average throughput enhancement for 5G FBMC with E-DSA Conclusion In this paper, an enhanced dynamic spectrum allocation (E-DSA) was developed by combining the FBMC 5G configuration with the cooperative game theory called the Nash bargaining solution (NBS) for spectrum allocation in order to improve 5G network throughput performance.
The FBMC transceiver was designed and simulated using the Labview communication software to obtain the BER values.
The S-BER values were then incorporated into the E-DSA algorithm to maximize the systemAos throughput.
The first result showed that the EDSA algorithm successfully helped in spectrum allocation among the primary user (PU) and the secondary users (SU.
where the bandwidth requests by the secondary users were supported by the bandwidthAos availability of the primary user.
Dynamic Spectrum Allocation Method DOI: 10.
5614/j.
The price weight factors were used to control and prevent the overpayment of costs of SUsAo bandwidthAos Secondly, for throughput maximization, it was proven that the throughputs for 5G can be improved tremendously with the implementation of E-DSA, where an increment of 140% of throughput was calculated theoretically and an increment of 144% was recorded from the simulation design.
The E-DSA results in this study were solely derived from simulations run on the Labview Communication software platform.
Future comparisons between the simulation findings and a real-time configuration should take into account the hardware implementation.
Adaptive characteristics, including the allotted subcarrier per channel quality need and the transform scheme, such as the wavelet transform, could also be examined.
Acknowledgement The authors acknowledge the Research Management Centre (RMC) of the International Islamic University Malaysia (IIUM) and the Malaysian Ministry of Education (MOHE) for the financial support for this research.
The research output is part of the deliverables for the research funded under IIUMAos Research University Initiatives.
The Fundamental Research Grant Scheme (FRGS) Research Project by the Malaysian Ministry of Education sponsors this research in the creation of novel theories and concepts.
References