Advertisement

Energy Saving in HetNet Network Using eNB Parameters Tuning

  • Narjes LassouedEmail author
  • Noureddine Boujnah
  • Ridha Bouallegue
Conference paper
Part of the Advances in Intelligent Systems and Computing book series (AISC, volume 927)

Abstract

Power consumption represents an exciting subject in the field of mobile communication networks. In the last decades, the radio access networks have witnessed a denoting augmentation in power consumption, and this has indirect drawbacks on both environment and economic sides. The increase of power consumption conduct to both a substantial growth in CO2 emissions and on the operational expenditure of network operators. For this reason, minimizing power consumption in mobile cellular networks has become mandatory. In this paper, we suggest a power saving scheme in LTE-Advanced based Heterogeneous Networks (HetNet) by the mean of switching ON/OFF eNodeB (eNB) and tilt modification of neighbors.

1 Introduction

Long Term Evolution (LTE), designed by 3GPP, is enhanced by the use of new access techniques and by the improvement of architecture that combines different new services. New methods are included to LTE to migrate to LTE Advanced such as Coordinated Multipoint (CoMP), Massive MIMO, carrier aggregation and HetNet. The notion of LTE Advanced based on HetNet aims to ameliorate spectral efficiency. HetNet is a combination between various cell types: microcells, picocells and femtocells, and separate access technology [1].

In recent years, data traffic in the wireless mobile network has gained considerable growth. Consequently, energy consumption of Information and Communication Technology (ICT) is rising with a staggering rate. This significant increase in ICT energy consumption has many environmental and economic effects. Thus, on the ecological side, ICT is responsible practically for 2% of CO2 emissions [2]. Moreover, on the economic side, the percentage of ICT energy consumption varies from 2% to 10% of the annual worldwide energy consumption [3].

Referring to Fig. 1 representing the consumption of power in a typical wireless mobile network [4], it is clear that the base station (BS) consume the biggest part of the energy that can attain 57 % among all networks [4]. Therefore, minimize energy consumption in the total network amounts to minimizing energy consumption in the BS.
Fig. 1.

Power consumption of a typical wireless cellular network

Several researches are proposed to reduce energy consumption in cellular networks, the most well-known are the Switch ON/OFF algorithms. The condition to apply the Switch ON/OFF procedure is different from one technique to another. The main aim is to reduce the energy consumption by optimizing the use of the power of BSs with guaranteed QoS.

In [5], according to the traffic variation, the authors switch ON/OFF dynamically the BSs by respecting certain conditions of blocking probability in order to guarantee the QoS. In [6], BSs are switched ON/OFF randomly to minimize energy consumption in a UMTS cellular network. An improvement of this work is given in [7] where authors proposed a Switch ON/OFF algorithm using a uniform and a hierarchical scenario.

Algorithms of Switch ON/OFF are also applied in LTE and LTE-A networks through some researches. Notably, in [8], Alexandra et al. try to come up with an optimum combination of switching ON/OFF eNBs in order to guarantee the maximum of energy saving. In [9], Narjes et al. implemented a Switch ON/OFF scheme in LTE network especially during low traffic period and QoS is guaranteed using a power up/down procedure. An improvement of this work is detailed in [10], where the algorithm of switch ON/OFF is implemented in an LTE-A based HetNet. In this work, the switch ON/OFF procedure is performed based on the SINR values of UEs. QoS and coverage are maintained using femtocells and CoMP techniques.

In all works mentioned above, researchers are concentrated on power consumption issue in UMTS, LTE and LTE-A networks. They all use the Switch ON/OFF procedure that aims to raise the power saving by minimizing the number of active eNBs during low traffic periods.

In this paper, we are interested by applying algorithms of Switch ON/OFF in LTE-A based HetNet. Our contribution is to switch off an eNB when the number of their attached UEs is lower than a given threshold. UEs of Switched off eNB are assigned after the Switch OFF depending on their measured SINR (Signal to Interference plus Noise Ratio) value. Femtocells will serve UEs with good SINR value and neighbors eNBs will serve UEs with bad SINR value. Coverage is maintained by applying a tilt variation procedure for the neighboring eNB.

The rest of the paper is organized as follows: Sect. 2, describes the Network architecture. Section 3, introduces the eNodeB power model. In Sect. 4, we describe the proposed algorithm. Results of simulation are the subject of Sects. 5 and 6 concludes the paper.

2 Network Architecture

For the network architecture, we propose a HetNet dense urban network, in which, neighboring cells overlaps. The considered topology is formed by N = 7 eNBs. Each eNB can serve one macrocell surrounded by two femtocells as shown in Fig. 2. For UEs generation, we apply a non uniform distribution of users where UEs are generated randomly between cells.
Fig. 2.

Regular network layout

In our topology, we focused on two main concepts used in LTE-Advanced.

2.1 Femtocells

Due to the use of HetNet, we benefit from the use of femtocells to reduce power consumption. At first, femtocells are developed to raise capacity and guarantee coverage for highly densified areas [1]. Moreover, Femtocells are small, inexpensive and are characterized by low energy consumption of BSs [11]. For this reason, in our work, femtocells are disposed around the eNB to compensate switched off procedure.

2.2 Antenna Tilt

The tilt represent the inclination or angle of the antenna to its axis. The tilt is used when we want to reduce interference and/or coverage of some specific areas [12]. In our case we will make tilt variation procedure in order to offer coverage to users of Switched OFF eNBs (Fig. 3).
Fig. 3.

Tilt presentation

3 Power Calculation Model

We will present, in this section, the power model used to calculate power saving. Starting by defining the eNB power model. At an instant t, the power of eNB i, \(P_{i}\) is given by:
$$\begin{aligned} P_{i}(t) = P_{OP} (i) + P_{TX}(i).S_i(t) \end{aligned}$$
(1)
Where, N is the total number of eNBs, \(P_{TX}\) is the total transmission power of eNB i, \(S_i(t)\) represent the eNB i state at an instant t and \(P_{OP}(i)\) is the eNB i operating power given by:
$$\begin{aligned} P_{OP}(i) = P_{a}(i) + P_{s}(i) + P_{c}(i) \end{aligned}$$
(2)
Where, \(P_a(i)\) is the eNB i power amplifier including Feeder, \(P_s(i)\) is the power supply for eNB i and \(P_c(i)\) is the cooling power consumed by eNB i. The transmitted power \(P_{TX}(i)\) is:
$$\begin{aligned} P_{TX}(i) = \sum _j^{N_u} P_{TX}(j) \end{aligned}$$
(3)
Where, \(N_u\) is the number of eNB i UEs and \(P_{TX}(j)\) is the power transmitted by UE j given by:
$$\begin{aligned} P_{TX}(j)=N_{RB}(j).P_{RB}(i) \end{aligned}$$
(4)
\(N_{RB}(j)\) is the number of resource blocks attributed to UE j and \(P_{RB}(i) \) is the power transmitted per resource block for eNB i. \(P_{RB}(i) \) is calculated using the following equation:
$$\begin{aligned} P_{RB}(i) =\frac{P_{TX}(i)}{ N_{RB}(i)} \end{aligned}$$
(5)
Where, \(P_{TX}(i) \) is the transmitted power defined in Eq. 3 and \(N_{RB}(i) \) is the total number of resource blocks attributed to eNB i. The state of eNB \(S_i(t)\) is defined as follows:
$$\begin{aligned} S_i(t) = \left\{ \begin{array}{rl} 1 &{} \text{ if } ON\\ 0 &{} \text{ if } OFF \end{array} \right. \end{aligned}$$
(6)
The total power of the network \(P_{total}\) at an instant t is given based on Eq. 1 by:
$$\begin{aligned} P_{total}(t) = \sum _i^{N} [P_{i}(t)] + \varDelta P = \sum _i^{N} [P_{OP} (i) + P_{TX}(i).S_i(t)] + \varDelta P \end{aligned}$$
(7)
Where, \(P_i\) is the eNB i power defined in Eq. 1 and \(\varDelta P\) is the amount of added power after applying the Switch ON/OFF algorithm.
The received power at UE’s side is calculated using the pathloss model defined in [13] as:
$$\begin{aligned} P_{RX}(\theta ) = P_{TX}.G(\theta ).D^{-\alpha } \end{aligned}$$
(8)
Where, \(P_{TX}\) is the eNB transmission power described in Eq. 4. D is the distance in km between the eNB and UE, \(\alpha \) is the propagation exponent selected depending on the type of propagation medium and \(G(\theta )\) is the antenna gain defined based on equations in [12] by:
$$\begin{aligned} G(\theta ) = G_{max}.10 ^{-12.(\frac{\arctan {(D/h)}-\theta }{\theta _0})^2} \end{aligned}$$
(9)
Where, \(G_{max}\) is the maximum antenna gain, D is the distance between UE and eNB, h is the antenna height, \(\theta _0\) is the elevation 3dB value, it may be assumed to be \(15^\circ \) and \(\theta \) is the angle tilt: \(\theta \in [ \arctan {(D/h)} - \theta _0.\sqrt{A_m/12}, .. ,\arctan {(D/h)} + \theta _0.\sqrt{A_m/12}]\), where \(A_m= 20 dB\) is the maximum attenuation. Let \(P_{min}\) be the power sensitivity of UE, the received power \(P_{RX}\) must be superior than \(P_{min}\).

4 Proposed Power Saving Algorithm

Our energy saving algorithm is based on the concept of Switch ON/OFF especially during low traffic periods when eNBs are little used. We add to this concept some new features that significantly reduce energy consumption in the network.

Our algorithm behaves as follow: At first, all macro eNBs are activated, and all femtocells are deactivated. When the number of active UEs attributed to an eNB i is smaller than a given threshold T, a Switch OFF procedure is applied to eNB i, at the same time, femtocells existing in the cell range of switched off eNB are activated to serve UEs surrounding the deactivated eNB. For UEs located at the cell extremity, they are allowed to the nearest active eNB neighbor.

Switching ON/OFF eNBs achieve a good amount of power saving but it affects the QoS by reducing the coverage; Thus, when an eNB is switched off, the coverage will be decreased and thereby, some UEs will be out of cell’s range. For this reason, we must compensate the coverage area to provide network coverage to UEs after switching off process. Thus, to achieve coverage continuity we will perform a power adjustment procedure combined with a tilt variation technique of the antenna tilt of neighbor eNBs that will serve the attached UE of the deactivated eNB. The Switch ON/OFF procedure is stopped when the eNB that we decide to switch OFF contains some active UEs existing in the cell extremity and there is no neighbor eNB in ON state. Steps for our proposed approach are:
  • Step 1: At first, we start by calculating the number of active UEs \(N_i(t)\) attributed to eNB i.

  • Step 2: The second step is to compare the value of calculated number \(N_i(t)\) to a fixed threshold T. We can meet two possible cases described in the algorithm below:
    Noting that \(S_f(t)\) is the state of the femtocell at an instant t given by:
    $$\begin{aligned} S_f(t) = \left\{ \begin{array}{rl} 1 &{} \text{ if } ON \\ 0 &{} \text{ if } OFF \end{array} \right. \end{aligned}$$
    (10)
  • Step 3: After taking the decision of Switch ON/OFF, we should specify new assignment of UEs of switched off eNB i. The allocation of a UE j is taken depending on its SINR value. The SINR of a UE j is calculated using the following formula:
    $$\begin{aligned} SINR(j)= \frac{P_{RX} (j)}{\sum _{i \ne j}^{N} P_{RX}(i)+ KT_pW} \end{aligned}$$
    (11)
    Where, \(P_{RX} (j)\) is the power received at UE j, given in Eq. 8, \( \sum _{i \ne j}^{N} P_{RX} (i)\) is the interference average power. \(KT_pW\) indicate the background noise, where K is the Boltzmann constant, \(T_p\) is the temperature and W is the bandwidth used by UE.
    The decision to allocate each UE to the correct cell is done referring to the Table 1.
    Table 1.

    SINR value according to UE position in LTE-A

    UE position

    SINR (dB)

    Around the eNB

    \(\ge \)13

    Middle of cell

    From 0 to 13

    End of cell

    \(\le \)0

    The best femtocell \(\widetilde{f}\) is chosen according to the following hypothesis:
    $$\begin{aligned} \widetilde{f} = \arg \max _{f \in N_{femtocells}} (P_{RX}^ {(f)} (i, j)) \end{aligned}$$
    (12)
    Where, \(N_{femtocells}(i)\) is the number of femtocells existing in the cell range of eNB i and \(P_{RX}\) is the received power at user side j coming from the femtocell f.
    The right eNB neighbor \(\widetilde{n}\) is chosen according to the following hypothesis:
    $$\begin{aligned} \widetilde{n} = \arg \max _{n \in N_{neighbors}} (P_{RX} ^{(n)}(i,j)) \end{aligned}$$
    (13)
    Where, \(N_{neighbors}(i) \) is the number of eNB i’s neighbors and \(P_{RX}\) is the power received at user side j coming from eNB neighbor n.
    When we activate femtocells and attribute to them UEs of deactivated eNB, a new amount of power denoted \(\delta P_{femto}\) is added to the total power \(P_{total}\). \(\delta P_{femto}\) is given by:
    $$\begin{aligned} \delta P_{femto}(t) =\sum _{f=1}^{N_{femto}} S_f (t).P_{f} \end{aligned}$$
    (14)
    \(P_{f} \) is the power of one femtocell and \(S_f(t)\) indicate its state defined in Eq. 10.
    The assignment of UEs is described in more details in the following algorithm:
  • Step 4: After allocation of UEs to femtocells, UEs in the cell borders are attributed to their nearest neighbor \(\widetilde{n}\). Two mechanisms are performed to ensure coverage: a power adjustment procedure is applied in parallel with a tilt variation of the antenna tilt of eNB neighbor \(n_0\). The antenna tilt variation’s process is performed in order to achieve coverage continuity and raise cells coverages.

    The better value of \(\theta \) should be chosen to offer coverage for UEs located in the cell extremity of deactivated eNB. Thus the choice of \(\theta \) should respect the following condition:
    $$\begin{aligned} P_{RX}(\theta ) < P_{min} \end{aligned}$$
    (15)
    If the previous condition is false, no tilt variation and no power addition is performed.
    When we apply the tilt variation procedure, an amount of power \(\delta P_{tilt}\) is added. \(\delta P_{tilt}\) depends on distance between the deactivated eNB and its neighbors:
    $$\begin{aligned} \delta P_{tilt}(t) = \max |P_{min} - P_{RX}(\theta )| \end{aligned}$$
    (16)
  • Step 5: After the procedure of Switch ON/OFF and power adjustment and tilt variation, the next step is to calculate the total power at instant t. \(\varDelta P \) in Eq. 7 is calculated as:
    $$\begin{aligned} \varDelta P (t) = \delta P_{femto} (t) + \delta P_{tilt} (t) \end{aligned}$$
    (17)
    Then, the total power \(P_{total} (t)\) at time t is given by:
    $$\begin{aligned} P_{total}(t) = \sum _i^{N} [ P_{OP} (i) + P_{TX}(i). S_i(t)]+ \sum _f^{N_{femto}} S_f (t). P_{f} + \sum _{k=1} ^ {M} [P_{min} - P_{RX}(\theta , k)] \end{aligned}$$
    (18)

5 Perfermance Evaluation

5.1 Simulation Scenario

To assess performances of our proposed scheme, we use an LTE-Advanced-based HetNet topology using MATLAB as implementation software. We propose an urban scenario where our topology is composed by \(N=7\) macrocells. The distance between eNBs is \(D=500\) m, and the eNBs-UEs distance is computed through their positions in the map. As shown in Fig. 4, in each macro cell we disposed two femtocells. At instant \(t=0\), all femtocells are in OFF state.
Fig. 4.

Network architecture

For the allocation of UEs between cells, we suppose a non-uniform and random distribution of UEs between cells where the maximum number of UEs served by an eNB is fixed to 100 UEs. The variation of the total number of active users during time is presented in Fig. 5.
Fig. 5.

Number of active UEs variation Vs time

For the simulation, we consider the eNB’s maximum transmit power for the downlink transmission that is recommended by [14]:
  • 43 dBm for a bandwidth of 1.25, 5 MHz.

  • 46/49 dBm for a bandwidth of 10, 20 MHz.

The simulation parameters employed in our simulation scenario are shown in Table 2.
Table 2.

Simulation parameters

Parameters

Values

Bandwidth [MHz]

20

Frequency [MHz]

2140

Propagation model

Pathloss model

Distribution of user

non uniform

Transmitted power for macro cell [dBm]

35–46

Transmitted power for femto cell [dBm]

5

Distance between macro eNBs [m]

500

Number of resource blocks (RBs) in macro cell

100

Number of resource blocks (RBs) in femto cell

20

5.2 Simulation Results

In this section, we will present simulation results and evaluation of the proposed scheme.

Starting by Fig. 6, representing the behavior of energy consumption in the network over time for different Switch ON/OFF algorithms. In this figure, we compare our proposed algorithm by several cases: the first case represented by the plot in blue describe the energy consumption without the use of any energy saving algorithm. The second curve in green represents the case of random Switch ON/OFF [6] where we chose to turn off 1/3 of BSs randomly. The curve in red describes the case of Switch ON/OFF at low traffic [9]. Finally, the pink plot represents the energy consumption after the use of our proposed Switch ON/OFF algorithm.
Fig. 6.

Power consumption Vs time

Looking at the Fig. 6, we notice that in high traffic hours, our proposed algorithm gives almost the same results as the low traffic algorithm, but in periods of low traffic, our proposed approach realizes the better power saving than the case of no switch ON/OFF technique is performed and the cases of low traffic and random schemes.

The percentage of power saving is the result of Fig. 7. As shown in this figure, the proposed scheme achieves the maximum amount of power saving than the random and the low traffic schemes. The percentage of power saving achieved by our proposed scheme can attain 32% especially during night period from 00:00 am to 05:00 am.
Fig. 7.

Power saving (%) Vs time

The use of femtocells improves the amount of energy saving; thus, femtocells are characterized by their low amount of energy consumption. Better yet, the use of the tilt variation technique increase the amount of power saving after applying the switch ON/OFF algorithm, thus, it improves the QoS by guarantying an acceptable coverage to UEs.

In order to evaluate QoS and coverage, we simulate SINR values before and after the Switch ON/OFF procedure. Using Eq. 11, we calculate SINR for UEs existing in the cell edge. The result of SINR calculation is presented in Fig. 8. As it is shown, the SINR values after the switch ON/OFF procedure increase and the coverage is guaranteed, Using the concept of femtocells and eNB parameters tuning let the QoS stable and not lot affected.
Fig. 8.

SINR values before and after the Switch ON/OFF procedure

6 Conclusion

The purpose of our work in this paper is to find a better method that can reduce energy consumption in a LTE-A based HetNet network. Our algorithm is based on two concept: the first one is the use of femtocells that can help to reduce the energy consumption in the network thanks to their low energy consumption. The second concept is the tilt variation technique that ensures an acceptable coverage and maintains QoS. According to the SINR values before and after Switch ON/OFF, we can say that our algorithm does not affect the QoS of the network.

Notes

Acknowledgment

This work has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No. 761579 (TERAPOD).

References

  1. 1.
    http://www.3gpp.org/hetnet. Accessed 15 Nov 2018
  2. 2.
    Global Action Plan Report, An inefficient truth, December 2017Google Scholar
  3. 3.
  4. 4.
    Han, C., et al.: Green radio: radio techniques to enable energy efficient wireless networks. IEEE Commun. Mag. 49(6), 46–54 (2011)CrossRefGoogle Scholar
  5. 5.
    Gong, J., Zhou, S., Niu, Z., Yang, P.: Traffic-aware base station sleeping in dense cellular networks. In: IWQoS 2018, pp. 1–2, June 2010Google Scholar
  6. 6.
    Chiaraviglio, L., Ciullo, D., Meo, M., Marsan, M.A.: Energy-aware UMTS access networks. In: W-GREEN 2008, Lapland (2008)Google Scholar
  7. 7.
    Chiaraviglio, L., Ciullo, D., Meo, M., Marsan, M.A.: Energy efficient management of UMTS access networks. In: ITC 21, Paris, France, pp. 1–8, September 2009Google Scholar
  8. 8.
    Bousia, A., Kartsakli, E., Alonso, L., Verikoukis, C.: Energy efficient base station maximization switch off scheme for LTE-advanced. In: IEEE International Workshop CAMAD, Barcelona, Spain, pp. 256–260, September 2012Google Scholar
  9. 9.
    Lassoued, N., Boujnah, N., Bouallegue, R.: Power saving approach in LTE using switching ON/OFF eNodeB and Power UP/DOWN of Neighbors, SETIT, Tunisia, December 2018Google Scholar
  10. 10.
    Lassoued, N., Boujnah, N., Bouallegue, R.: Reducing power consumption in HetNet network using power adjustment and coordinated multipoint technique. In: AINA, Cracow, Poland, pp. 539–544, May 2018Google Scholar
  11. 11.
    Andrews, J.G., Claussen, H., Dohler, M., Rangan, S., Reed, M.C.: Femtocells: past, present, and future. IEEE J. Sel. Areas Commun. 30(3), 497–508 (2012)CrossRefGoogle Scholar
  12. 12.
    Guidelines for evaluation of radio interface technologies for IMT-Advanced. Radiocommunication Sector of ITU (2009)Google Scholar
  13. 13.
    Salous, S.: Radio Propagation Measurement and Channel Modeling. Wiley, New York (2013)CrossRefGoogle Scholar
  14. 14.
    E-UTRA; further advancements for E-UTRA physical layer aspects, 3GPP (2010)Google Scholar

Copyright information

© Springer Nature Switzerland AG 2019

Authors and Affiliations

  • Narjes Lassoued
    • 1
    • 2
    Email author
  • Noureddine Boujnah
    • 3
    • 4
  • Ridha Bouallegue
    • 1
  1. 1.SUP’COM, INNOV’COM Research Lab., Higher School of Communication of TunisUniversity of CarthageTunisTunisia
  2. 2.National Engineering School of Gabes, ENIGGabesTunisia
  3. 3.Faculty of Science of Gabes, FSGGabesTunisia
  4. 4.Waterford Institute of Technology, TSSGWaterfordIreland

Personalised recommendations