Professional Documents
Culture Documents
Research Article
Abstract: Low-power wide-area network (LPWAN) is an important category of internet of things network topology. LPWAN
connects a massive number of low-cost devices with low-power consumption and long coverage range. Narrow band internet of
things (NB-IoT) is a Third Generation Partnership Project (3GPP) standard specifically designed for LPWAN. Owing to massive
connectivity random access channel (RACH) of NB-IoT often gets overloaded. This potential problem needs to be addressed
immediately. For controlling the RACH overload, 3GPP has introduced two access barring mechanisms in LTE, which are (i)
access class barring and (ii) extended access barring (EAB). In NB-IoT only EAB mechanism has been adopted. In this study,
the authors show through simulations that broadcasting of the access barring factor and barring time in system information
block-2 can be used to improve the success probability and reduce the access delay of random access mechanism of NB-IoT
for different traffic models.
Nomenclature (EAB), (iii) separate resources for MTC, (iv) dynamic allocation of
RA channel (RACH) resources, (iv) back-off specific schemes, and
N total number of UEs in the cell (v) slotted access and pull-based RA [3]. There are also non-3GPP
δi periodicity of NPRACH/slot length in CE level I based solutions for dealing with congestion in IoT. Notable
Ni initial number of UEs in CE level I schemes include self-optimisation overload control RA [4],
Si number of sub-carriers allocated to CE level I prioritised RA [5], group-based RA [6], spatial group-based
M maximum number of attempts in all CE levels reusable preamble allocation [7], a reliability guaranteed RA [8],
Ri maximum number of attempts per preamble allowed in CE non-aloha-based RA [9], and collision-resolution-based RA [10].
level I 3GPP has adapted ACB and EAB for LTE/LTE-A. However, in
Wi size of back-off windows in CE level I NB-IoT only EAB mechanism has been accommodated.
α access barring factor The objective of this paper is to demonstrate the benefit of ACB
T acb access barring time mechanism in improving the RA procedure of NB-IoT as
Ps probability of successful access compared to the traditional back-off method. We carry out the
D mean delay of the successfully accessed UEs extensive simulation on different traffic models and demonstrate
that broadcasting access barring (AB) factor and barring time
information in system information block-2 (SIB-2) can improve the
1 Introduction access probability and reduce the access delay of RA mechanism
Recently, internet of things (IoT) has attracted great attention due of NB-IoT.
to the advancement of wireless technologies. Basic idea is to Rest of the paper is organised as follows, Section 2 explains the
connect a wide variety of physical devices through the internet. An RA procedure in NB-IoT. The AB mechanisms are presented in
important class of IoT networks is the low-power wide-area Section 3. Simulation results on the effect of access bearing
network (LPWAN). LPWAN connects devices, that have stringent mechanism on the access probability and access delay of NB-IoT
requirements on battery life and cost, over the wide area [1, 2]. are presented in Section 4. Finally Section 5 concludes the paper.
Some of the promising IoT applications based on LPWAN include
automated energy metre reading, smart grid, smart homes etc. [1]. 2 Random access procedure
Narrow band IoT (NB-IoT) is the standard introduced by Third
Generation Partnership Project (3GPP) in release 13 specifically RA procedure starts only after the UE has obtained downlink
for LPWAN [1, 2]. NB-IoT is similar to LTE with differences in synchronisation, decoded master information block (MIB) and had
radio structure, for adapting to low bandwidth. In NB-IoT, user valid SIBs. In NB-IoT cell is divided into three coverage
equipment (UE) can talk to E-UTRAN Node B (eNB), the base enhancement levels (CE levels). CE level is decided by the UE by
station, only after uplink synchronisation. Random access (RA) is a comparing received power with the received signal received power
crucial step for UE in synchronising with the network. There are (RSRP) thresholds broadcasted by the eNB in the SIB-2. SIB-2
two types of RA procedures namely contention-based and also carries information pertaining to RA configuration for three
contention-free procedures. In the contention-free RA procedure, CE levels. RA in NB-IoT involves four steps as outlined below
the eNB allocates resources to UE for accessing the network. In the [11].
contention-based RA procedure devices compete for the resources
to access the network. NB-IoT follows contention-based RA 2.1 Preamble transmission (Msg1)
procedure. RA procedure in NB-IoT is similar to LTE. However,
The first step, involves the transmission of Msg1 or preamble in
NB-IoT deals with huge number devices and hence there will be
the narrow-band physical RACH (NPRACH). NPRACH is a set of
severe congestion.
48 sub-carriers which are divided into four sets of 12 each. Those
Many solutions have been proposed by 3GPP to deal with
four sets of 12 sub-carriers are mapped to the three CE levels.
congestion reduction in LTE/LTE-A. The proposed schemes
Unlike in LTE, the preamble in NB-IoT is based on sub-carrier and
include (i) access class barring (ACB), (ii) extended access barring
IET Wirel. Sens. Syst., 2019, Vol. 9 Iss. 6, pp. 399-404 399
© The Institution of Engineering and Technology 2019
20436394, 2019, 6, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-wss.2019.0036 by Cochrane Luxembourg, Wiley Online Library on [04/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
the hopping pattern [12]. The preamble in NB-IoT is a set of four 3.2 Extended access barring
symbol-groups with each symbol-group transmitted in different
sub-carrier as per the hopping pattern. Each symbol-group consists In EAB UEs are divided into ten classes numbered from 0 to 9
of one cyclic prefix plus five symbols. The preamble is transmitted based on the number stored in their subscriber identity module.
by randomly selecting the initial sub-carrier. Further preamble There are other classes 11–15 which are allocated for emergency
transmission is repeated several times. After preamble cases. In EAB, a bitmap is broadcasted in SIB-14 with 10 bits. The
transmission, UE starts RA response (RAR) timer. UEs belonging to class with bit indicated as one can access the
network while UEs belonging to other classes are barred from
accessing the network. EAB mechanism is activated only in case of
2.2 Random access response (Msg2) high traffic load. EAB activation is indicated by a bit in SIB-14.
After preamble is received and decoded at the eNB, the eNB sends Only when EAB activation bit is set to one, the bitmap is
the RARAR that includes the RA preamble identifier, timing broadcasted.
advance, temporary cell radio network identity (TC-RNTI) and There are many works analysing ACB and EAB in LTE. In
some scheduling messages. UE expects the RAR reception before [14], the authors presented an analytical model for the EAB
RAR timer expiration. If RAR is not received before RAR timeout algorithm used in LTE. In [15], the authors explained the limiting
the UE re-transmits the preamble after waiting for the back-off factors of EAB algorithm. While the impact of barring rates and
interval. Due to the presence of multiple CE levels, UEs from barring time of ACB on network performance has been studied in
lower CE levels can transmit preamble in higher CE level. For [16]. Few works have appeared recently in the literature comparing
adapting to transmission in higher CE levels, a new counter has EAB and ACB. In [17], the authors showed that optimal
been introduced in addition to the original counter. The original performance of ACB is better than EAB in terms of energy
counter keeps track of the total number of preamble transmission consumption.
attempts made by a UE in all CE levels. The new counter counts NB-IoT system adopted only the EAB mechanism and the UEs
the preamble attempts in the current CE level, once the maximum in NB-IoT system reads SIB-14 only when the AB bit is set to one
preamble attempts of the current CE level is reached before the in MIB [13]. During bursty arrivals, if AB bit is set to 0 then
original counter reaches the maximum value then the new counter collision happens before the activation of EAB. Even if EAB is
resets to zero and UE considers itself to be in the next CE level and activated in case there are a large number of UEs belonging to the
continues with the contention. This process continues until either same class then the problem of congestion remains. In this paper,
UE moves to the next step or all preamble attempts are exhausted. we show that the combination of ACB with back-off method
RA is considered to be a failure once the original counter reaches improves access probability.
the maximum number of attempts.
4 Simulation and comparison
2.3 RRC connection request (Msg3) In order to study the effect of ACB on the RA mechanism of NB-
Upon receiving the RAR, UE transmits the TC-RNTI and some IoT we carried out an extensive simulation. Simulations were
MAC layer information in case of data transmission, on the performed using C++ program executing on the linux platform. For
scheduled resources and starts contention resolution timer. the purpose of simulation, the RA algorithm of NB-IoT has been
modelled as multi-band multi-channel slotted ALOHA as in [18].
2.4 RRC connection setup (Msg4) An ideal physical layer has been assumed to isolate the effect of the
physical layer on the RA performance. UE re-transmits the
In the last step, eNB decodes the TC-RNTI and assigns a preamble only when the collision happens due to initial sub-carrier
permanent C-RNTI, which acts as an UE identifier for further selection. Notations used in our study are given in Nomenclature
communication. All UEs that have selected the same initial sub- section.
carrier in preamble transmission will receive the same RAR and
they send the TC-RNTI through the same scheduled resources. 4.1 Configuration of parameters and traffic models
Since eNB cannot decode multiple UEs, a collision occurs.
Collided UEs re-transmits the preamble. Even if Msg4 is not The values of parameters used in the simulation of RA procedure
received within contention resolution window UE re-transmits the of NB-IoT are set to be the same as in [18]. We have assumed that
preamble. the parameters of the ACB to be static, that is AB factor and AB
time does not change with load. Simulations are done for the above
parameters for N UEs which access the network and each UE
3 AB mechanisms generate a single access request during a period of time T i.e. all
AB Mechanism is used for regulating the access request traffic by devices activate between t = 0 and t = T. The RA intensity is
barring some UEs from accessing the network for certain interval described by distribution p(t). The number of UEs which generate
of time. ACB and EAB are the AB mechanisms adopted by 3GPP their access requests in the ith access opportunity is defined as
[13]. Both mechanisms are used in LTE/LTE-A. follows
ti + 1
3.1 Access class barring
In ACB, a barring factor (α) and a barring time (T acb) are
A(i) = N ∫
ti
p(t) dt
1
p(t) = for a ≤ t ≤ b
b−a
400 IET Wirel. Sens. Syst., 2019, Vol. 9 Iss. 6, pp. 399-404
© The Institution of Engineering and Technology 2019
20436394, 2019, 6, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-wss.2019.0036 by Cochrane Luxembourg, Wiley Online Library on [04/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
we consider the packets to N UEs arrive uniformly over first 5 s. simulation results for certain configuration of NB-IoT against the
corresponding analytical or simulation results reported already in
(iii) Beta distribution is a continuous probability distribution the literature. Since ACB was not a recommended access bearing
defined on the interval [0,1] with two positive shape parameters α mechanism in NB-IoT standard, no analytical model has been
and β and the pdf p(t) is defined as follows: reported in the literature for modelling the performance of the NB-
IoT with ACB mechanism. The closest model that can be used for
tα − 1(1 − t) β − 1 validation of our results is that of Harwahyu et al. [18]. Harwahyu
p(t) = for α, β > 0 et al. [18] reports analytical and simulation results for the
B(α, β)
probability of successful access and the mean delay experienced by
a UE of NB-IoT, for the case of normal backoff mechanism (i.e.
Packets arrive to N UEs in the first 5 s based on Beta distribution
without access bearing factor and bearing time). In order to carry
with shape parameters α = 3, and β = 4.
out the above-mentioned comparison for the normal backoff
One-shot traffic occurs during group paging, uniform traffic is
condition we carried out our simulation as follows. We ran our C+
considered as a realistic scenario and beta traffic is considered as
+ simulation code by setting the access bearing factor α = 1 (please
an extreme scenario [3]. For simulations, N is varied from 60 to
refer Section 3.1). The value of access bearing time is immaterial if
1200 in steps of 60.
α = 1, but we have set T barring = 0.
In Figs. 1 and 2, we compare our simulation results for NB-IoT
4.2 Validation of our simulation procedure with normal backoff (α = 1) against the analytical results reported
We first validate the correctness of our simulation procedure before by the authors of [18] for probability of successful access and mean
presenting the results. To this end, we first try to compare our delay experienced by the UEs. The network configurations of NB-
Fig. 1 Probability of successful access: comparison of our simulation results for α = 1 with one shot traffic model and network configuration given by
Table 1 against the analytical results of [18]. N denotes the number of UEs
Fig. 2 Mean delay of the UEs: comparison of our simulation results for α = 1 with one shot traffic model and network configuration given by Table 1 against
the analytical results of [18]. N denotes the number of UEs
IET Wirel. Sens. Syst., 2019, Vol. 9 Iss. 6, pp. 399-404 401
© The Institution of Engineering and Technology 2019
20436394, 2019, 6, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-wss.2019.0036 by Cochrane Luxembourg, Wiley Online Library on [04/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Fig. 3 Probability of successful access by UEs against the number of UEs Fig. 5 Probability of successful access by UEs against the number of UEs
in ‘One-shot’ model. Network configuration is given by Table 1 in ‘Uniform’ model. Network configuration is given by Table 1
402 IET Wirel. Sens. Syst., 2019, Vol. 9 Iss. 6, pp. 399-404
© The Institution of Engineering and Technology 2019
20436394, 2019, 6, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-wss.2019.0036 by Cochrane Luxembourg, Wiley Online Library on [04/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Fig. 7 Probability of successful access by of UEs against the number of UEs in ‘Beta’ model. Network configuration is given by Table 1
Fig. 8 Average packet transmision delay of successful UEs against the number of UEs in ‘Beta’ model. Network configuration is given by Table 1
Table 2 Ps and D for different sets of α and T acb for Beta traffic model for N = 1500
α T acb = 2 s T acb = 4 s T acb = 8 s T acb = 16 s
Ps D Ps D Ps D Ps D
0.5 0.71 8.07 0.82 14.74 0.98 22.21 0.999 29.008
0.7 0.57 4.90 0.69 7.85 0.79 14.49 .95 23.49
0.9 0.40 2.83 0.49 4.25 0.57 7.05 0.64 13.17
5 Conclusion GLOBECOM Workshops (GC Wkshps), Houston, USA, December 2011, pp.
368–372
In this paper, we have demonstrated the benefit of using ACB for [7] Jang, H.S., Kim, S.M., Ko, K.S., et al.: ‘Spatial group based random access
for M2M communications’, IEEE Commun. Lett., 2014, 18, (6), pp. 961–964
better performance of RA mechanism of NB-IoT. We showed the [8] Madue no, G.C., Pratas, N., Stefanovic, C., et al.: ‘Massive M2M access with
trade-off between the probability of successful access and mean reliability guarantees in LTE systems’. 2015 IEEE Int. Conf. on
packet transmission time. We further showed how this trade-off is Communications (ICC), London, UK, 2015, pp. 2997–3002
affected by the parameters of ACB namely AB time and AB factor. [9] Shirvanimoghaddam, M., Li, Y., Dohler, M., et al.: ‘Probabilistic rateless
multiple access for machine-to-machine communication’, IEEE Trans. Wirel.
Optimal adjustment and/or dynamic adjustment of AB factor and Commun., 2015, 14, (12), pp. 6815–6826
barring time are possible future extensions to the reported work. [10] Ali, M.S., Hossain, E., Kim, D.I.: ‘LTE/LTE-a random access for massive
machine-type communications in smart cities’, IEEE Commun. Mag., 2017,
55, (1), pp. 76–83
6 References [11] 3GPP: ‘Evolved universal terrestrial radio access (EUTRA); medium access
[1] Ericsson and A.B: ‘Cellular networks for massive IoT enabling low power control (MAC) protocol specification’, Technical Specification (TS) 36.321,
wide area applications’, January 2016, pp. 1–13 version 13.2.0, June 2016
[2] GSMA: ‘3GPP low power wide area technologies white paper’, October [12] Lin, X., Adhikary, A., Wang, Y.P.E.: ‘Random access preamble design and
2017, pp. 1–49 detection for 3GPP narrowband IoT systems’, IEEE Wirel. Commun. Lett.,
[3] 3GPP: ‘Study on RAN improvements for machine-type communications’, 2016, 5, (6), pp. 640–643
Technical Specification (TR) 37.868, version 11.0.0, September 2011 [13] 3GPP: ‘Evolved universal terrestrial radio access (EUTRA); radio resource
[4] Yilmaz, O.N.C., Hmlinen, J., Hmlinen, S.: ‘Self-optimization of random control (RRC)’, Technical Specification (TS) 36.331, version 13.2.0, June
access channel in 3GPP LTE’. 2011 7th Int. Wireless Communications and 2016
Mobile Computing Conf., Istanbul, Turkey, July 2011, pp. 1397–1401 [14] Cheng, R.G., Chen, J., Chen, D.W., et al.: ‘Modeling and analysis of an
[5] Lo, A., Law, Y.W., Jacobsson, M., et al.: ‘Enhanced LTE-advanced random- extended access barring algorithm for machine-type communications in LTE-
access mechanism for massive machine-to-machine (M2M) communications’. a networks’, IEEE Trans. Wirel. Commun., 2015, 14, (6), pp. 2956–2968
27th World Wireless Research Forum (WWRF) Meeting (WWRF27- [15] Zhang, Z., Chao, H., Wang, W., et al.: ‘Performance analysis and UE-side
WG4-08), Atlantic City, USA, October 2011, pp. 1–5 improvement of extended access barring for machine type communications in
[6] Cheng, J.P., Lee, C.H., Lin, T.M.: ‘Prioritized random access with dynamic LTE’. 2014 IEEE 79th Vehicular Technology Conf. (VTC Spring), Seoul,
access barring for RAN overload in 3GPP LTE-a networks’. 2011 IEEE South Korea, May 2014, pp. 1–5
IET Wirel. Sens. Syst., 2019, Vol. 9 Iss. 6, pp. 399-404 403
© The Institution of Engineering and Technology 2019
20436394, 2019, 6, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-wss.2019.0036 by Cochrane Luxembourg, Wiley Online Library on [04/02/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
[16] Leyva-Mayorga, I., Tello-Oquendo, L., Pla, V., et al.: ‘Performance analysis [18] Harwahyu, R., Cheng, R.G., Wei, C.H., et al.: ‘Optimization of random access
of access class barring for handling massive M2M traffic in LTE-a networks’. channel in NB-IoT’, IEEE Internet Things J., 2018, 5, (1), pp. 391–402
2016 IEEE Int. Conf. on Communications (ICC), Kuala Lumpur, Malaysia, [19] Wei, C., Cheng, R., Tsao, S.: ‘Modeling and estimation of one-shot random
May 2016, pp. 1–6 access for finite-user multichannel slotted aloha systems’, IEEE Commun.
[17] Wali, P.K., Das, D.: ‘Optimization of barring factor enabled extended access Lett., 2012, 16, (8), pp. 1196–1199
barring for energy efficiency in LTE-advanced base station’, IEEE Trans.
Green Commun. Netw., 2018, 2, (3), pp. 830–843
404 IET Wirel. Sens. Syst., 2019, Vol. 9 Iss. 6, pp. 399-404
© The Institution of Engineering and Technology 2019