Heterogeneous networks spectral efficiency analysis with modified time-domain interference coordination algorithm in various load distribution scenarios for 5G New radio Текст научной статьи по специальности «Медицинские технологии»

HETEROGENEOUS NETWORKS SPECTRAL EFFICIENCY ANALYSIS WITH MODIFIED TIME-DOMAIN INTERFERENCE COORDINATION ALGORITHM IN VARIOUS LOAD DISTRIBUTION SCENARIOS

FOR 5G NEW RADIO

DOI 10.24411/2072-8735-2018-10334

Roman V. Glazkov,

The Bonch-Bruevich Saint-Petersburg State University of Telecomminications, Saint-Petersburg, Russia; University of Jyvaskyla, Finland, roman.v.glazkov@gmail.com

Alexandra V. Nikitina,

The Bonch-Bruevich Saint-Petersburg State University

of Telecomminications, Saint-Petersburg, Russia, Keywords: 5G New Radio, Spectral Efficiency,

envision@yandex.ru Interference Coordination, HetNet, ABS.

The provision of required normalized throughput on cell edges and areas with insufficient signal strength can be achieved by heterogeneous networks establishment. The key problem of these systems is a cross-tier interference between macrocells and smallcells which should be diminished. One of the common methods to cope with the channel resource shortage is the load balancing. In homogeneous networks users connect to the base station providing the maximum SINR level. In HetNets it leads to a traffic disbalance as the cells with weak transmit power (picocell and femtocell) are chosen as serving cells only with very close range UEs or not chosen at all. In such situation Cell Range Expression (CRE) technique is used. The main idea of CRE is to artificially increase SINR level from weak cells by several dB (3 - 4 dB for LTE Release 10 and above) [2]. Terminals in range expansion zone experience low downlink SINR and control channels failure may occur. Therefore a total spectral efficiency increase task becomes a challenging problem.

The aim of this paper is to estimate cell-edge spectral efficiency in heterogeneous networks (HetNets) reducing cross-tier interference with Almost Blank Subframes (ABS) for 5G New Radio applications. Two situations of user traffic are considered: both macro- and small cells are overloaded or only small cell is underloaded. To detect ABS density in each of the described scenarios modified time-domain Interference Coordination algorithm is used [1]. This solution is designed to improve overall system performance evaluating the number of victim users and optimal ABS density by means of utility function maximization. The network model was built in ns-3 discrete event network framework. Cumulative distribution functions (CDFs) of normalized throughput are calculated in various load distribution cases. The results obtained within the proposed approach are given and discussed.

Для цитирования:

Глазков Р.В., Никитина А.В. Анализ спектральной эффективности гетерогенных сетей с модифицированным во временной области алгоритмом координации интерференции в различных сценариях распределения нагрузки для сетей 5G New Radio // T-Comm: Телекоммуникации и транспорт. 2019. Том 13. №12. С. 56-61.

For citation:

Glazkov R.V., Nikitina A.V. (2019). Heterogeneous networks spectral efficiency analysis with modified time-domain interference coordination algorithm in various load distribution scenarios for 5G New Radio. T-Comm, vol. 13, no.12, pр. 56-61.

1 Introduction

Enhanced ICIC approach (Interference Coordination) was developed to protect small cell control channels by means of resource coordination in frequency and time domains. For these purposes in LTE Release 10 Carrier Aggregation and Almost Blank Subframes were introduced. With a further development and newer releases the orthogonal frequency division multiplexing symbol shift was presented. Macro-femto and macro-pico interference coordination had several advanced designs of ABS-based eiCtC including self-optimized designs with regard to key parameters such as ABS muting ratio and joint optimized designs of ABS-based elCIC and other radio resource management techniques such as user association and power control [3]. In this work wc consider the ABS-based elCiC approach and its application for spectral efficiency improvements.

With the time-domain ICIC approach the network coordination system schedule resources in the time domain and provide for victim users special Almost Blank Subframes without data and control information. In most publications the ABS density is chosen with empirical methods. These studies suppose that number of ABS is approximately equal to the number of all pico- or femto- users |4-fi|. Other papers consider that ABS density parameter can be expressed via suggested formulas [7] and with a usage of the heterogeneous key performance indicators [8]. Nevertheless these approaches don't take into account the precise amount of victim users so obtained results might be optimized for dynamic allocation of users.

In this paper modified Interference Coordination algorithm was studied with the aim to find an optimal ABS density parameter value in four various load distribution situations: only macrocell is overloaded, only femtocell is underloaded, both macro and femtoceUs are overloaded, all the cells are underloaded. An important option of the proposed algorithm is a dynamic calculation of the real victims number in accordance to the scenario.

In the first part of this article the modified Interference Coordination algorithm is described. In the second part model parameters are expressed. The last part contains simulation results and conclusions.

2 Calculation of ABS density parameter

The modified time-domain Interference Coordination algorithm determines the optimal value of factor alpha that is equal to the number of Almost Blank Subframes in bitmap {ABS density parameter). Optimization carries out while maximum cell-edge spectral efficiency at cell edges is being reached. The key feature in this case is two-staged "victim" UE detection. At the first stage S1NR threshold criterion was set (-80 dBm in our scenario) and tracking of all UE was effected. The presence of users with low signal's level was defined. When such users appeared, calculation of factor alpha was launched by means of the algorithm and ABS mode was chosen automatically. Mentioned procedure is depicted in Fig. 2. After this is only the "victim" UF group was considered and then each femtocell reselection of "victim" UEs was realized. At the end, optimal value of factor alpha was obtained.

Also it's possible to carry out real time calculations because of the fact that computational complexity is reduced from 0(2 ) to 0(K:). In this case "2" is the number of kits taking part in se-

lection ("victim" and "normal"), K is total amount of small cell's users.

(1/10,1) ¡0000000001, ...j

H ■¡■II 1 1:1

Fig. 1. Example of a bitmap frame with factor alpha is equal to 0.1

Track UE and victim detection based on 5!NR threshold • Activation of ABS mode H SINR tracking duringABS DcrctlwäonofABSmod«

Fig. 2. ABS mode control

Net parameters UE selection based on SINR level ABS mode

initialization activation

i r

Tracking of UE's Optimal factor UE reselection

SINR level alpha calculation

ABS mode

deactivation

Fig. 3. Proposed algorithm

Modified algorithm is represented in Fig. 3 step by step. The main idea of involved algorithm at the first point is to maximize the utility function of "normal" and "victim" users given for each value of a that can be determined as:

Here

= max

H

Jm-\,n\a)+ log

(\-a)R

j (m-irl-C(mr

m'" G(mf

(1)

RL = G{NUa)) rilNUcc)t R> =

are the achievable data rate of "normal" and "victim" users re-

N'r.UO 1 Nj,Aa) |

spec lively; G(NJP+(a)) = £ —; G(NJp_(a)) = £ -;

PiI » SI n

rm , rn are average data rate of "normal" and "victim" users respectively; m,n are the amounts of "normal" and "victim" users respectively.

At the same time each small cell determine it's own number of "victim" UE. Dynamic programming is applied during reselection as represented in Fig. 4.

At the second stage each macrocell maximize it's own utility function also given for each possible a and expressed as:

£/;w («) = log(l - a) ■ N'M + ^ log R¡ Here R;=G(N:yl)-r;/NM.

7TT

Input data

I ri. iVirtue l.Jt}

V"

Ni =0

Initialization

(0.01a) = ^'(0.01 a) =

frtKT) * = î to E for {Vnt.H« 0.-t) if

C/*„,(fn,n|a) = max-

im -l)"н■G(m)',,

■<Hn)" -I)"'

Output data

U*_ „(a)=max{üj__ (m,n ja) \m+ n = K)

Fig. 4. Dynamic programming

The last step eonsist of two operations. Firstly the utility functions are maximized (see formula 3) of macrocells and small cells from previous calculations and derive an array of common utility functions for each given u.

U(a) = VM(a)+UF(a) -£e/," (a) +

M

c(«) (3)

f=l

Program initialization end introduction of model variables

Introduction of virtual model area where u Es and eNB are located

ABS mode is used

Decision about ABS mode launching

Output infcmaiion : measured specific transmission cata rate values end SINR values

Introduction of UE distribution model and pathless model

eNBdisWt virtual moc and UE to connection utionin ü area eNB

ASS mode isn't used

Fig. 5. Imitation model

Decision about ABS mode usage depends on measured SINR values. Statistic measurements every 4-5 seconds are an important feature because of handover procedure. Namely, those UE who have low SINR values while crossing cell borders arc able to pass them.

3 ITU channel mode)

The channel model ITU-R P. 1411-6 [9] and a pan of which appropriated for urban conditions is used in our model. Path loss calculations are represented below:

L -L

LoSJ bp

f d 1

20 lOg, g KRn> j f d 1

40loglu

for d < R.

for d > R

bp

where L. is the lowest value of path loss;

* IT

where Rbp is the threshold distance value between a cell and UB;

(5)

(6)

( ,1 \

25 log, , CI R, V *P J / \

401og,„ (i)

for d < R/

for d > R.

O)

Here UM(a), UF(a) are the utility functions calculated on

macrocells and femtocells respectively.

Then algorithm scans this array and choose the final common function. îndex a that corresponds this chosen function is optimal ABS density parameter and the solution of our optimization problem. Limitations during optimization process are the following:

0.1 < a < 0.91 m + n-K (4>

Here K is the number of users taking part in reseleetion.

Time configuration information between maerocell and femtocell during data exchange in ABS mode translates across commutation center as a bitmap pattern of 4 frames (40 bits). The same way is used when calculated values of utility function is translated.

Algorithm verification is implemented in imitation model created in ns-3 simulator. This model contains I maerocell and 1 femtocell. Corresponding scheme is presented in Fig. 5.

where ¿.j. is the highest value of path loss;

12 ^

Lbp ~

20 log

10

%xhhhm

(8)

where L is the basic transmission loss; hh - height of Base

Station antenna; hm height of UE's antenna; X - carrier's

wave lenght.

Total pass loss are defined as:

L=LLOS,I+LLOS,U

(9)

Allowed pass loss defined as:

^=-S + G + Pbs

(10)

where Pbs = 46 dUm transmission power of the Base Station;

S = -85 dBm - UE's receiver sensitivity; G — 14,5 dBm - transmitter's antenna gain.

^=-S + G+P,, = H6dBm (II)

A = -

Taking into account that hh - 30 m, hm — 1.5 m and f ,

where c - speed of light, a/- carrier frequency we obtain Rbj>

according to (6). So R. te 1273 m. We consider the distance

between the Base Stations is equal to 1000 m therefore d, which is the distance between UE and Base Station, don't exceed 1000 m and d < Rhp . Then = 92.9 dB according to (7),

L,„s„ -112,4 dB according to (8). Finally, using these results in

(9) we have L = 102,7 dB, Obtained value is lower than the allowed pass loss value so current configuration doesn't suffer from significant interference during signal propagation.

T

Table 2

Simulation results (case 2)

Before algorithm execution Average spectral efficiency, bit/s/Hz Cell-edge spectra! efficiency, bit/s/Hz Typical modification at the cell edge

All UE 0.211 0.063

Macrocell UE 0.205 0.061

Femtocell UE 0.213 0.066

After algorithm execution

All UE 0.134 0,044 Increased at 1.43 times

Macrocell UE 0.085 0,016 Decreased at 3.81 times

Femtocell UE 0.378 0,112 Increased at 1.69 times

Numerical values of average and cell — edge spectral efficiency measured in corresponding points of graphics are brought to the tables 1 and 2,

Conclutions

The simulation model of 1 macrocell and I femtocell was examined. Average and cell spectral efficiency in several scenarios were compared. In situation when both cells are overloaded increasing average and cell-edge spectra) efficiency of femtocell users were observed at 1.69 and 1.77 times correspondingly and at the same time reduction of this parameter in macrocell UE terminals (at 3.81 and 2.41 times correspondingly).

It was supposed that the factor alpha is equal to 0.4 to be an optimal value for this simulation. It's evident that the higher alpha values utilized the smaller time spans intended to macrocell service. As a result a total macrocell spectral efficiency UE diminishes.

When speaking about femtocell overloading (case 2) it is worth mentioning that proposed algorithm also improve average and cell-edge spectral efficiency of femtocell UE (at 6.95 and 6.816 times correspondingly). Optimal factor alpha is choscn to be 0.6 and the macrocell UE suffer from decreasing spectral efficiency as in the first situation. The difference is as follows: when macrocell has low user distribution density the spectral efficiency reduction effect is not so strong.

Taking into account all the above it is possible to conclude that modified Inter Cell Interference Coordination algorithm

shows robust efficiency when one can deal with critical amount of UE in the cell. The further work direction could be done for 5G and Beyond special case scenarios as unsupervised and highly loaded heterogeneous networks with fast moving nodes and higher spectral efficiency and throughput requirements.

References

1. Pang J., Wang J., Wang D., Shen G. (2012). Optimized time domain resource partitioning for enchanccd inter-cell interference coordination in heterogeneous network. IEEE Wireless Communications am1 Networking Conference: Mac and cross layer design. 5p,

2. Ezzaouia Mahdi, Gueguen C., Yassin, Mohamad, Am mar Malimoud, Lagrange Xavier, Bouallcgue Л. (2017). Autonomous and dynamic inter-cell interference coordination techniques for future wireless networks. 1-8. 10.110WWiMOB.20I7.8115759.

3. Ling, Liu & Zhou, Yiqing & Vasilakos, Athanasios & Tian, Lin & Shi, Jinglin. (2019). Time-domain ICIC and optimized designs for 5G and beyond: a survey. Science China Information Sciences. 62. 10.1007/sl 1432-017-9477-4.

4. Wang Y., Pederson K.l. (20) 1). Time and Power Domain Interference Management for LTE Networks with Macro-Cells and 1-leNBs. IEEE Proc. Vehic. Tech. Conf. Sept. 2011. 7 p.

5. 3GPP R1 - N2331. (2011). Performance evaluation in heterogeneous networks. 43 p.

6. 3GPP R1 - 112411. (2011). Scenarios for further enhanced noil ca-bascd ICIC for LTE. 56 p.

7. El Shaer H. (2012). Interference management scheme for heterogeneous network with cell range extension in Network Operations and Management Symposium (APNOMS). Degree project in signal processing. Stockholm, Sweden. 76 p.

8. Alvarez, Beatriz Soret; De Do mem со, Antonio; Bazzi, Samer; Malimood, Nurul Hilda;Pedersen, Klaus I. (2018). Interference Coordination for 5G New Radio. IEEE Wireless Communications Magazine.

9. Recommendation ITU-R P.I41I-6 (2012), "Propagation data and prediction methods for the planning of short-range outdoor radio communication systems and radio local area networks in the frequency range 300 MHz to 100 GHz".

10. REPORT ITU-R M.2I34 (2008), "Requirements related to technical performance for lMT-Advanced radio interfaee(s)"

11. 3GPP TR 38.201 Release 15 (2019), "NR; Physical layer; General description"

12. Narcis Cardona, Jose F. Monserrat, Jorge Cabrejas (2012). Enabling Technologies for 3GPP LTE-Advanced Networks. Universitat Polit'ecnica de Vaf cncia, Spain, pp.1 -33.

т

АНАЛИЗ СПЕКТРАЛЬНОЙ ЭФФЕКТИВНОСТИ ГЕТЕРОГЕННЫХ СЕТЕЙ С МОДИФИЦИРОВАННЫМ ВО ВРЕМЕННОЙ ОБЛАСТИ АЛГОРИТМОМ КООРДИНАЦИИ ИНТЕРФЕРЕНЦИИ В РАЗЛИЧНЫХ СЦЕНАРИЯХ РАСПРЕДЕЛЕНИЯ НАГРУЗКИ ДЛЯ СЕТЕЙ 5G NEW RADIO

Глазков Роман Викторович, Санкт-Петербургский государственный университет телекоммуникаций им. проф. М.А.Бонч-Бруевича, г. Санкт-Петербург, Россия; Университет города Jyvaskyla, Финляндия (University ofJyvaskyla), roman.v.glazkov@gmail.com

Никитина Александра Викторовна, Санкт-Петербургский Государственный Университет Телекоммуникаций им проф.М.А. Бонч-Бруевича, г. Санкт-Петербург, Россия, envision@yandex.ru

Аннотация

Целью данной работы является оценка спектральной эффективности на границе сот в гетерогенных сетях (HetNets) для снижения межсотовой интерференции с помощью использования так называемых "почти пустых субкадров" (Almost Blank Subframes - ABS) в сетях нового поколения 5G New Radio. Рассматриваются две ситуации пользовательского трафика в макро- и фемто- сотах: перегружены как макро так и фемто-соты, либо часть сот не испытывает перегрузки по трафику. Для определения плотности ABS в каждом из описанных сценариев используется алгоритм координации интерференции, модифицированный во временной области [1]. Это решение предназначено для улучшения общей производительности системы путем оценки количества обслуживаемых абонентов, выявления абонентских групп с недостаточным качеством обслуживания и обеспечения оптимальной плотности ABS с помощью максимизации функции полезности. Модель сети была построена с помощью симулятора ns-3 при использовании дискретного режима моделирования сетей. Рассчитаны кумулятивные функции распределения (CDF) нормализованной пропускной способности в различных случаях распределения нагрузки. Приведены и рассмотрены результаты, полученные в рамках предложенного подхода.

Ключевые слова: 5G New Radio, спектральная эффективность, алгоритм координации интерференции, гетерогенные сети.

Литература

1. Pang J., Wang J., Wang D., Shen G. Optimized time domain resource partitioning for enchanced inter-cell interference coordination in heterogeneous network" IEEE Wireless Communications and Networking Conference: Mac and cross - layer design, 2012. С. 1613-1617

2. Ezzaouia Mahdi, Gueguen C., Yassin Mohamad, Ammar Mahmoud, Lagrange Xavier, Bouallegue A. Autonomous and dynamic inter-cell interference coordination techniques for future wireless networks." I0.II09/WiMOB.20l7.8ll5759, 2017. С 1-8.

3. Ling Liu, Zhou Yiqing, Vasilakos Athanasios, Tian Lin, Shi Jinglin. Time-domain ICIC and optimized designs for 5G and beyond: a survey. Science China Information Sciences. 62. I0.I007/sII432-0I7-9477-4, 2019. 62 с.

4. Wang Y., Pederson K.I. Time and Power Domain Interference Management for LTE Networks with Macro-Cells and HeNBs", IEEE Proc. Vehic. Tech. Conf., 20II, 7 с.

5. 3GPP RI - II233I, (20II), "Performance evaluation in heterogeneous networks", 43 с.

6. 3GPP RI - II24II, (20II), "Scenarios for further enhanced non ca-based ICIC for LTE", 56 с.

7. El Shaer H. (20I2), "Interference management scheme for heterogeneous network with cell range extension in Network Operations and Management Symposium (APNOMS)", Degree project in signal processing, Стокгольм, Швеция, 76 с.

8. Alvarez Beatriz Soret, De Domenico Antonio, Bazzi Samer, Mahmood Nurul Huda, Pedersen Klaus I. Interference Coordination for 5G New Radio // IEEE Wireless Communications Magazine, 20I8. С. I3I-I37.

9. ITU-R P.I4II-6 (20I2), "Propagation data and prediction methods for the planning of short-range outdoor radio communication systems and radio local area networks in the frequency range 300 MHz to I00 GHz", 35 с.

10. ITU-R M.2I34 (2008), "Requirements related to technical performance for IMT-Advanced radio interface(s)", 8 с.

11. 3GPP TR 38.20I Release I5 (20I9), "NR; Physical layer; General description", I4 с.

12. Narcis Cardona, Jose F. Monserrat, Jorge Cabrejas. Enabling Technologies for 3GPP LTE-Advanced Networks. Universitat Polit'ecnica de Val'encia, Испания, 20I2. С. I-33.

Информация об авторах:

Глазков Роман Викторович, Сотрудник научно-образовательного центра "Технологии информационных и образовательных систем" Санкт-Петербургского государственного университета телекоммуникаций им. проф. МА.Бонч-Бруевича, г. Санкт-Петербург, Россия Аспирант университета города Jyvaskyla, Финляндия (University of Jyvaskyla)

Никитина Александра Викторовна, Доцент кафедры Радиосвязи и Вещания Санкт-Петербургского Государственного Университета Телекоммуникаций им проф.МА. Бонч-Бруевича, г. Санкт-Петербург, Россия

( I л