Indonesian J our nal of Electrical Engineering and Computer Science V ol. 23., No. 1, July 2021, pp. 308 320 ISSN: 2502-4752, DOI: 10.11591/ijeecs.v23.i1.pp308-320 r 308 NB-IoT and L TE-M to wards massi v e MTC: Complete perf ormance e v aluation f or 5G mMTC Adil Abou El Hassan, Abdelmalek El Mehdi, Mohammed Saber Mohammed First Uni v ersity Oujda, National School of Applied Sciences, SmartICT Lab, Oujda, Morocco Article Inf o Article history: Recei v ed Oct 29, 2020 Re vised May 1, 2021 Accepted May 15, 2021 K eyw ords: 3GPP 5G IoT L TE-M mMTC NB-IoT Performance e v aluation ABSTRA CT Since the emer ging 5G wireless netw ork is e xpected to significantly re v olutionize the field of communication, its standardizati on and design should re g ard the internet of things (IoT) among the main orientations. Also, emer ging IoT applications introduce ne w requirements other than throughput to support massi v e machine-type commu- nication (mMTC) where small data pack ets are occasionally sent. Therefore, more importance is attached to co v erage, latenc y , po wer consumption, and connection den- sity . F or this purpose, the third generation partnership project (3GPP) has introduced tw o no v el cellular IoT technologies supporting mMTC, kno wn as NB-IoT and L TE- M. This paper aims to determine the system configuration and deplo yment required for NB-IoT and L TE-M technologies to fully meet the 5G mMTC requirements i n terms of co v erage, throughput, latenc y , battery life, and connection density . An o v ervie w of these technologies and their design principles are also described. A complete e v alua- tion of NB-IoT and L TE-M performance ag ainst 5G mMTC requirements is presented, and it is sho wn that these requirements can be met b ut only under certain conditions re g arding system configuration and deplo yment. This is follo wed by a performance comparati v e analysis , which is mainly conducted to determine the limits and suitable use cases of each technology . This is an open access article under the CC BY -SA license . Corresponding A uthor: Adil Abou El Hassan Mohammed First Uni v ersity Oujda National School of Applied Sciences SmartICT Lab, Oujda, Morocco Email: a.abouelhassan@ump.ac.ma 1. INTR ODUCTION Internet of things (IoT) is seen as a dri ving force behind recent impro v ements in wireless communica- tion technologies such as third generation partnership project (3GPP), long term e v olution adv anced (L TE-A) and 5G ne w radio (NR) to meet the e xpected requirements of v arious massi v e machine-type communication (mMTC) applications. The mM TC introduces a ne w communication era where billions of de vices, such as remote indoor or outdoor sensors, will need to communicate t ogether while being connected to the cloud-based system. The purpose of 5G system design is to co v er three cate gories of use cases: enhanced mobile broadband (eMBB), massi v e m achine-type communication (mMTC), as well as ultra reliable lo w latenc y communication (uRLLC) which is designed to support critical machine-type communication (cMTC) [1]. The adv antage of the 5G s ystem is the fle xibility of its structure, which allo ws the use of a common inte grated system to co v er man y use cases, by using a ne w feature which is netw ork slicing based on softw are-defined netw orking (SDN) J ournal homepage: http://ijeecs.iaescor e .com Evaluation Warning : The document was created with Spire.PDF for Python.
Indonesian J Elec Eng & Comp Sci ISSN: 2502-4752 r 309 and netw ork function virtualization (NFV) technologies [2]. There are currently man y lo w po wer wide area (LPW A) technologies such as SigF ox and LoRa, b ut their deplo yment requires ne w infra structure implementation with no benefit from e xisting L TE system. There- fore, 3GPP has introduced in Release 13 (Rel-13) specifications tw o LPW A technologies for IoT : narro wband IoT (NB-IoT) and L TE machine-type communication (L TE-M(TC)) [3]. The 3GPP Rel-13 core specifications for NB-IoT and L TE-M were finalized in June 2016 [4], [5], while Rel-14 and Rel-15 enhancements were completed i n June 2017 and J une 2018 re specti v ely [4], [5]. As for Rel-16 enhancements, the y were completed in June 2020 whereas Rel-17 enhancements are underw ay and scheduled for completion in June 2022 [1]. The 3GPP design aims for Rel-13 were lo w cost and lo w comple xity de vices, long battery life, and co v erage enhancement to reaching NB-IoT and L TE-M de vices in poor co v erage conditions. F or this purpose, tw o po wer sa ving techniques ha v e been implemented t o reduce de vice po wer consumption: po wer sa ving mode (PSM) and e xtended discontinuous reception (eDRX) introduced in Rel-12 and Rel-13 respecti v ely [3], [6]. About Rel-15, 3GPP has defined in its w ork v e requirements of 5G mMTC in terms of co v erage, throughput, latenc y , battery life, and connection density [7]. Man y papers address 3GPP LPW A technologies including NB-IoT and L TE-M and non-3GPP LPW A technologies such as LoRa and Sigfox. El Soussi et al. [8] implement NB-IoT and L TE-M modules in net- w ork simulator NS-3, to e v aluate only battery life, latenc y , and connection density . Whereas J ¨ ork e et al. [9] e v aluate only throughput, latenc y , and battery life of NB-IoT and L TE-M. Ho we v er , Liber g et al. [10] focus on NB-IoT technology only b ut pro vide a performance e v aluation ag ainst 5G mMTC requirements. On the other hand, Krug et al. [11] compare the delay and ener gy consumption of data transfer co v ering v arious IoT communication technologies such as Bluetooth, W iFi, LoRa, Sigfox, and NB-IoT . Ho we v er , to our kno wledge, there is no paper co v ering the performance e v aluation of the NB-IoT and L TE-M technologies ag ainst the v e requirements of 5G mMTC as well as the comparati v e analysis of these performances. This paper aims to determine the system configuration and deplo yment required for NB-IoT and L TE- M technologies to fully meet the 5G mMTC requirements. Our contrib ution is to perform a comparati v e analysis of the performances of NB-IoT and L TE-M technologies, based on the e v aluated performances ag ainst the 5G mMTC requir ements to determine the limits and suitable use cases of each technology . The remainder of the paper is or g anized as follo ws. An o v ervie w of the NB- IoT and L TE-M technologies is pro vided in Section 2. The performance e v aluation methodology of NB-IoT and L TE-M technologies is presented in Section 3. This is follo wed, in Section 4, by a complete performance e v aluation of NB-IoT and L TE-M technologies ag ainst the 5G mMTC requirements i n terms of co v erage, throughput, latenc y , battery life, and connection density . Afterw ard, a comparati v e analysis of the e v aluated performances of NB-IoT and L TE-M technologies is presented. Also, the enhancements pro vided by the recent 3GPP releases are discussed. Finally , Section 5 concludes the paper . 2. O VER VIEW OF CELLULAR IO T TECHNOLOGIES: NB-IO T AND L TE-M 2.1. Narr o wband IoT : NB-IoT The bandwidth occupied by the NB-IoT carrier is 180 kHz corresponding to one ph ysical resource block (PRB) of 12 subcarriers in an L TE system [12]. There are three operation modes to deplo y NB-IoT : as a stand-alone carrier , in guard-band of an L TE carrier and in-band within an L TE carrier [13]. T o coe xist with the L TE system, NB-IoT uses orthogonal frequenc y di vision multiple access (OFDMA) in do wnlink with an identical subcarrier spacing of 15 kHz and frame structure as L TE [14]. Whereas NB-IoT uses in uplink single- carrier frequenc y di vision multiple access (SC-FDMA) and tw o transmission modes which are the multi-tone and single-tone transmissions to ensure both high capacity and maximum co v erage for NB-IoT de vice with a single antenna [14]. Multi-tone transmission uses the sam e 15 kHz subcarrier spacing and 0 .5 ms slot duration as L TE, while single-tone transmission supports tw o numerologies that use 15 kHz and 3.75 kHz subcarrier spacings with 0.5 ms and 2 ms slot durations respecti v ely [15]. The restricted quadrature phase-shift k e ying (QPSK) and binary phase-shift k e ying (BPSK) modulation schemes are used in both do wnlink and uplink [16]. Also, NB-IoT defines three co v erage enhancement (CE) le v els in a cell: CE-0, CE-1, and CE-2 corresponding to the maximum coupling loss (MCL) of 144 dB, 154 dB, and 164 dB respecti v ely [17]. T w o de vice cate gories Cat-NB1 and Cat-NB2 are defined by NB-IoT which correspond to the de vice cate gories introduced in Rel-13 and Rel-14 respecti v ely . The maximum transport block size (TBS) supported in uplink by Cat-NB1 is only 1000 bits compared to 2536 bits for Cat-NB2. Whereas for do wnlink, the maximum NB-IoT and L TE-M towar ds massive MTC: Complete performance e valuation ... (Adil Abou El Hassan) Evaluation Warning : The document was created with Spire.PDF for Python.
310 r ISSN: 2502-4752 TBS supported by Cat-NB1 is only 680 bits compared to 2536 bits for Cat-NB2 [4]. The signals and channels used in do wnlink (DL) are as follo ws: narro wband primary synchronization signal (NPSS), narro wband secondary synchronization signal (NSSS), narro wband reference signal (NRS), nar - ro wband ph ysical broadcast channel (NPBCH), narro wband ph ysical do wnlink shared channel (NPDSCH) and narro wband ph ysical do wnlink control channel (NPDCCH) [12], [16]. NPDCCH is used to transmit do wnlink control information (DCI) for uplink, do wnlink and paging scheduling [12], [16]. In the uplink (UL), only one signal and tw o channels are used: demodulat ion reference s ignal (DMRS), narro wband ph ysical uplink shared channel (NPUSCH) and narro wband ph ysical random access channel (NPRA CH). T w o formats are used for NPUSCH which are: F ormat 1 (F1) and F ormat 2 (F2). NPUSCH F1 is used by the user equipment (UE) to carry uplink user’ s data to the e v olv ed node B (eNB), and it supports both single-t one and multi-tone transmis- sions [17]. Whereas NPUSCH F2 is used to carry uplink control information (UCI), such as h ybrid automated repeat request-ackno wledgement (HARQ-A CK) and it supports only single-tone transmission [17]. F or cell access, the UE must first synchronize with the eNB using NPSS and NSSS signals to achie v e time and frequenc y synchronization with the netw ork and cell identification. Then, it recei v es narro wband mas- ter information block (MIB-NB) and narro wband system information block 1 (SIB1-NB) carried by NPBCH and NPDSCH respecti v ely from eNB to access the system [12], [16]. 2.2. L TE-machine (type communication): L TE-M(TC) L TE-M reuses an identical frame structure and also the same numerology as L TE, OFDMA is used in do wnlink while SC-FDMA is used in uplink with a subcarrier spacing of 15 kHz in both uplink and do wnlink [18], [19]. The L TE-M transmissions are limited to a narro wband si ze of 6 PRBs correspondings to 1.4 MHz including guardbands [ 3 ] . As the L TE system has a bandwidth from 1.4 to 20 MHz, some non-o v erlapping narro wbands (NBs) can be used if the L TE bandwidth e xceeds 1.4 MHz [20]. Up to Rel-14, L TE-M de vice uses QPSK and 16-QAM modulation schemes with a single antenna for both do wnlink and uplink. Whereas the support of 64-QAM in do wnlink has been introduced in Rel-15 [20]. T w o de vice cate gories are defined by L TE-M: Cat-M1 and Cat-M2 corresponding to de vice cate gories introduced in Rel-13 and Rel-14 respecti v ely . Cat-M1 has only a maximum channel bandwidth of 1.4 MHz compared to 5 MHz for Cat-M2 [20]. Besides, Cat-M2 supports a lar ger TBS of 6968 bits and 4008 bits in uplink and do wnlink respecti v ely , compared to 2984 bits in both do wnlink and uplink for Cat-M1 [5]. The follo wing channels and signals are reused by L TE-M in DL: ph ysical do wnlink shared channel (PDSCH), ph ysical broadcast channel (PBCH), pri mary synchronization signal (PSS), secondary synchroniza- tion signal (SSS), positioning reference signal (PRS), and cell-specific reference signal (CRS). MTC ph ysical do wnlink control channel (MPDCCH) is the ne w control channel that has the role of carrying DCI for uplink, do wnlink and paging scheduling [5], [19]. whereas for UL, the follo wing signals and channels are reused: demodulation reference signal (DMRS), sounding reference signal (SRS), ph ysical uplink shared channel (PUSCH), ph ysical random access channel (PRA CH), and ph ysical uplink control channel (PUCCH) which con v e ys UCI [5], [19]. F or cell access, the UE uses the PSS/SSS signals to synchronize with the eNB. Then it uses PBCH which carries the master information block (MIB), and PDSCH which con v e ys the ne w system information block 1 for reduced bandwidth UEs (SIB1-BR) from eNB to access the system [19]. 3. METHODOLOGY OF NB-IO T AND L TE-M PERFORMANCE EV ALU A TION The methodology used to perform a complete performance e v aluation for both NB-IoT and L TE-M technologies is based on the li nk le v el si mulations (LLS) as part of 3GPP’ s w orks using Ericsson’ s adv anced simulation tool [21–24]. The e v aluated performances correspond to the v e requirements of 5G mMTC in terms of co v erage defined by the MCL, throughput, latenc y , battery life, and connection density . 3.1. Co v erage The MCL is a common measure to define the le v el of co v erage a system can support. It is depending on the maximum transmitter po wer ( P T X ) , the required signal-to-interference-and-noise ratio (SINR), the recei v er noise figure (NF), and the signal bandwidth (BW) [25]: M C L = P T X ( S I N R + N F + N 0 + 10 l og 10 ( B W )) (1) Indonesian J Elec Eng & Comp Sci, V ol. 23., No. 1, July 2021 : 308 320 Evaluation Warning : The document was created with Spire.PDF for Python.
Indonesian J Elec Eng & Comp Sci ISSN: 2502-4752 r 311 where N 0 is the thermal noise density which is a constant equal to -174 dBm/Hz and P T X is referred to as either transmission po wer per carri er of eNB for do wnlink MCL calculation or the transmission po wer of UE for uplink MCL calculation. Whereas the SINR v alue comes from the LLS and depends on the tar geted block error rate (BLER) associated with each channel. 3.2. Thr oughput The do wnlink and uplink throughputs of NB-IoT are obtained according to the NPDSCH and NPUSCH F1 transmission time interv als issued from NPDSCH and NPUSCH F1 scheduling c ycles respecti v ely and that are pro vided by the LLS. While the do wnlink and uplink throughputs of L TE-M are determined based on the PDSCH and PUSCH transmission time interv als issued from PDSCH and PUSCH scheduling c ycles respec- ti v ely and also pro vided by the LLS. The MA C-layer throughput (THP) is calculated as follo ws: T H P = (1 B LE R )( T B S O H ) P D C C H P er iod (2) It is note w orth y that the TBS of the ph ysical do wnlink shared channel is used for the do wnlink THP calculation, whereas the TBS of the ph ysical uplink shared channel is used for the uplink THP calculation. While OH denotes the o v erhead size in bits corresponding to the radio protocol stack. Kno wing that the peri- odicity of the user -specific search spaces of ph ysical do wnlink control channel T is defined by the product of the relati v e starting subframe periodicity (G) and the maximum number of repetitions ( R max ): T = G R max [17], [26]. Therefore, the PDCCH period is a multiple of T which corresponds to the periodicity of the sched- uled transmissions of ph ysical do wnlink and uplink shared channels that are used for do wnlink and uplink THP calculation respecti v ely . 3.3. Latency The latenc y is defined as the delay between the de vice synchronization to the cell and the deli v ery of a data pack et to the eNB. It should be e v aluated for the follo wing procedures: radio resource control (RRC) Resume procedure and early data transmission (EDT) procedure that has been introduced in Rel-15 and al- lo wing the de vice to terminate the t ransmission of small data pack ets earlier in RRC-idle mode. Figure 1 (a) and Figure 1 (b) depict the data and signaling flo ws corresponding to the RRC Resume and EDT procedures used by NB-IoT respecti v ely . While the data and signaling flo ws corresponding to the RRC Resume and EDT procedures used by L TE-M are illustrated in Figure 2 (a) and Figure 2 (b) respecti v ely . The pack et definitions and their sizes used for the latenc y e v aluation of NB-IoT and L TE-M at the MCL of 164 dB are gi v en in T able 1 according to [21]. As sho wn in Figure 1 (a) and Figure 2 (a), the data pack et in RRC Resume procedure is transmit ted to the eNB together with the Message 5. Whereas in EDT procedure, the data pack et is transmitted to the eNB together with the Message 3 as sho wn in Figure 1 (b) and Figure 2 (b). (a) (b) Figure 1. Data and signaling flo ws for NB-IoT latenc y e v aluation; (a) RRC resume procedure and (b) EDT procedure NB-IoT and L TE-M towar ds massive MTC: Complete performance e valuation ... (Adil Abou El Hassan) Evaluation Warning : The document was created with Spire.PDF for Python.
312 r ISSN: 2502-4752 (a) (b) Figure 2. Data and signaling flo ws for L TE-M latenc y e v aluation; (a) RRC resume procedure and (b) EDT procedure T able 1. P ack et’ s definitions and sizes for latenc y e v aluation of NB-IoT and L TE-M RRC Resume procedure EDT procedure Random Access Response (Msg2) 7 bytes Random Access Response (Msg2) 7 bytes RRC Conn. Resume Request (Msg3) 11 * j 7 ** bytes RRC Conn. Resume Request (Msg3) + UL report 11 + 105 bytes RRC Conn. Resume (Msg4) 19 bytes RRC Conn. Release (Msg4) 24 * j 25 ** bytes RRC Conn. Resume Complete (Msg5) 22 + 200 bytes + RLC Ack Msg4 + UL report RRC Conn. Release 17 * j 18 ** bytes * P ack et size of NB-IoT ** P ack et size of L TE-M 3.4. Battery life The RRC resume procedure is used for battery life e v aluation instead of the EDT procedure since EDT procedure does not support uplink TBS lar ger than 1000 bits which requires long transmission times. The pack et flo ws used to e v aluate battery life of NB-IoT and L TE-M are t he same as sho wn in Figure 1 (a) and Figure 2 (a) respecti v ely , where DL data corresponds to the application ackno wledgment re g arding UL report receipt by eNB. F our le v els of de vice po wer consumption are defined, including transmission ( P T x ), reception ( P R x ), Idle-Light sleep ( P I LS ) corresponding to the de vice in RRC-Idle mode or RRC-Connected mode b ut not acti v ely recei ving or transmitting, whereas Idle-Deep sleep ( P I D S ) corresponds to po wer sa ving mode. The battery life in years is calculated using the follo wing formula according to [27]: B atter y l if e [ y ear s ] = B atter y ener g y capacity 365 E day 3600 (3) where E day is the de vice ener gy consumed per day in Joule and calculated as (4) E day = [( P T x T T x + P R x T R x + P I LS T I LS ) N r ep ] + ( P I D S 3600 24) (4) T T x , T R x and T I LS correspond to o v erall times gi v en in seconds for transmission, reception, and Idle-Light sleep respecti v ely according to pack et flo ws sho wn in Figure 1 (a) and Figure 2 (a) and obtained from the transmission times of signals and do wnlink and uplink channels pro vided by the LLS, while N r ep corresponds to the number of uplink reports per day . 3.5. Connection density The 5G mMTC tar get on connection density that is also part of the International Mobile T elecom- munication tar gets for 2020 and be yond (IMT -2020), requires the support of one million de vices per square kilometer in four dif ferent urban macro scenarios [7]. These scenarios are based on tw o channel models (UMA A) and (UMA B) and tw o distances of 500 and 1732 meters between adjacent ce ll sites denoted by ISD (inter - site distance) [28]. Based on the simulation assumptions gi v en in T able 2 and the non-full b uf fer system le v el Indonesian J Elec Eng & Comp Sci, V ol. 23., No. 1, July 2021 : 308 320 Evaluation Warning : The document was created with Spire.PDF for Python.
Indonesian J Elec Eng & Comp Sci ISSN: 2502-4752 r 313 simulation to e v aluate connection density of NB-IoT and L TE-M according to [22], Figure 3 sho ws the latenc y required at 99% reliability to deli v er 32 bytes of payload as a function of the connection requests intensity (CRI) to be supported, corresponding to the number of de vice’ s connection requests per second, cell and PRB. T able 2. System le v el simulation assumptions of urban macro scenarios P arameter V alue Frequenc y band 700 MHz L TE and L TE-M system bandwidths 10 MHz - 1.4 MHz Operation mode of NB-IoT In-band P athloss model UMA A, UMA B eNB po wer and antennas configuration 46 dBm - 2Tx/2Rx De vice po wer and antennas configuration 23 dBm - 1Tx/1Rx Figure 3. Intensity of connection requests in relation to latenc y The latenc y sho wn in Figure 3 is e v aluated by using the RRC Resume procedure and includes the idl e mode time of the de vice to synchronize to the cell and read the MIB-NB/MIB and SIB1-NB/SIB1-BR. Kno wing that each de vice must submit a connection request to the system periodically , we can calculate the connection density to be supported (CDS) per cell area using the follo wing formula: C D S = C R I C R P A (5) where CRP is the periodicit y of the de vice’ s connection requests gi v en in seconds and the he xagonal cell area A is calculated as follo ws: A = I S D 2 p 3 = 6 . 4. PERFORMANCE EV ALU A TION RESUL TS AND DISCUSSION 4.1. Ev aluation of NB-IoT and L TE-M perf ormance 4.1.1. Co v erage The simulation assumptions and system model paramet ers used to e v aluate the do wnlink and uplink MCL are gi v en in T able 3 according to [21]. Based on the simulation assumptions and using (1) to calculate MCL, T able 4 and T able 5 sho w the NB-IoT and L TE-M channel co v erage respecti v ely , to achie v e the MCL of 164 dB which corresponds to the 5G mMTC co v erage requirement to be supported [7]. NB-IoT and L TE-M towar ds massive MTC: Complete performance e valuation ... (Adil Abou El Hassan) Evaluation Warning : The document was created with Spire.PDF for Python.
314 r ISSN: 2502-4752 T able 3. Simulation and system model parameters P arameter V alue L TE and L TE-M system bandwidths 10 MHz - 1.4 MHz Channel model / Doppler spread T apped Delay Line (TDL-iii NLOS) / 2 Hz NB-IoT mode of operation Guard-band eNB po wer and antennas configuration 46 dBm - 4Rx/2Tx and 4Rx/4Tx for only (N)PSS/(N)SSS transmissions De vice po wer and antennas configuration 23 dBm - 1Rx/1Tx T able 4 and T able 5 also indicate the required acquisition time and block error rate (BLER) associated with each channel to achie v e the tar geted MCL of 164 dB. From the acquisition times sho wn in T able 4 and T able 5, we note that to reach the MCL of 164 dB at the appropriate BLER, it is mandatory to use the time repetition technique for the simulated channels. T able 4. Do wnlink and uplink co v erage of NB-IoT Assumptions Do wnlink ph ysical channel Uplink ph ysical channel for simulation NPBCH NPDCCH NPDSCH NPRA CH NPUSCH F1 NPUSCH F2 TBS [Bits] 24 23 680 - 1000 1 Acquisition time [ms] 1280 512 1280 205 2048 32 BLER 10% 1% 10% 1% 10% 1% Max transmit po wer [dBm] 46 46 46 23 23 23 T ransmit po wer/carrier [dBm] 35 35 35 23 23 23 Noise figure NF [dB] 7 7 7 5 5 5 Channel bandwidth [kHz] 180 180 180 3.75 15 15 SINR [dB] -14.5 -16.7 -14.7 -8.5 -13.8 -13.8 MCL [dB] 163.95 166.15 164.15 164.76 164 164 T able 5. Do wnlink and uplink co v erage of L TE-M Assumptions Do wnlink ph ysical channel Uplink ph ysical channel for simulation PBCH MPDCCH PDSCH PRA CH PUSCH PUCCH TBS [Bits] 24 18 328 - 712 1 Aquisition time [ms] 800 256 768 64 1536 64 BLER 10% 1% 2% 1% 2% 1% Max transmit po wer [dBm] 46 46 46 23 23 23 T ransmit po wer/carrier [dBm] 39.2 36.8 36.8 23 23 23 Noise figure NF [dB] 7 7 7 5 5 5 Channel bandwidth [kHz] 945 1080 1080 1048.75 30 180 SINR [dB] -17.5 -20.8 -20.5 -32.9 -16.8 -26 MCL [dB] 163.95 164.27 163.97 164.7 164 165.45 4.1.2. Thr oughput Figure 4 depicts NPDSCH scheduling c ycle of NB-IoT according to [21], where the NPDCCH user - specific search space is configured wi th a maximum repetition f actor R max of 512 and a relati v e starting subframe periodicity G of 4. Whereas the NP USCH F1 scheduling c ycle depicted in Figure 5 corresponds to the scheduling of NPUSCH F1 transmission once e v ery fourth scheduling c ycle according to [21]. Figure 4. NPDSCH scheduling c ycle ( R max =512 ; G=4) at the MCL Indonesian J Elec Eng & Comp Sci, V ol. 23., No. 1, July 2021 : 308 320 Evaluation Warning : The document was created with Spire.PDF for Python.
Indonesian J Elec Eng & Comp Sci ISSN: 2502-4752 r 315 Figure 5. NPUSCH F1 scheduling c ycle ( R max =512 ; G=1.5) at the MCL Based on BLER and TBS gi v en in T able 4 and using an o v erhead (OH) of 5 bytes according to [21], the MA C-layer THP is 281 bps on both do wnlink and uplink according to the (2). Figure 6 depicts the PDSCH scheduling c ycle of L TE-M which corresponds to the scheduling of PDSCH transmission once e v ery third scheduling c ycle, where the MPDCCH user -specific search space is configured with R max of 256 and a relati v e starting subframe periodicity G of 1.5 according to [21]. Whereas the PUSCH scheduling c ycle depicted in Figure 7 corresponds to the scheduling of PUSCH transmission once e v ery fifth scheduling c ycle according to [21]. Figure 6. PDSCH scheduling c ycle ( R max =256 ; G=1.5) at the MCL Figure 7. PUSCH scheduling c ycle ( R max =256 ; G=1.5) at the MCL From BLER and TBS indicated in T able 5 and the use of an o v erhead (OH) of 5 bytes, the M A C -layer throughputs obtained in do wnlink and uplink are 245 bps and 343 bps respecti v ely according to the (2). As part of 3GPP Rel-15, 5G mMTC requires that do wnlink and uplink troughputs supported at the MCL of 164 dB must be at least 160 bps [7]. As can be seen, the MA C-layer throughputs of both NB-IoT and L TE-M technologies meet the 5G mMTC requirement, which corresponds to the suitable throughput for IoT applications using sporadic transmissions of small data pack ets. It should be noted that the BLER tar gets associated with each NB-IoT and L TE-M channel require the acquisition times sho wn in T able 4 and T able 5 respecti v ely . Therefore, the throughput le v els can be further impro v ed by using the ne w de vice cate gories Cat-NB2 and Cat-M2 which support a lar ger TBS in both do wnlink and uplink with enhanced HARQ processes. 4.1.3. Latency The latenc y e v aluation is based on the same system model with the parameters gi v en in T able 3 and using the simulation assumptions relating to the RR C Resume and EDT procedures indicated in T able 1. Using the RRC Resume procedure, the e v aluated latenc y of NB-IoT is 9 seconds, while the EDT procedure allo ws obtaining a latenc y of only 5.8 seconds according to [21]. Re g arding the latenc y e v aluation of L TE- M, the latencies obtained by using the RRC Resume and EDT procedures are 7.7 and 5 seconds respecti v ely . Therefore, the 5G mMTC tar get of 10 seconds latenc y at the MCL of 164 dB defined in 3GPP Rel-15 [7] is met by NB-IoT and L TE-M technologies for both RRC Resume and EDT procedures. Ho we v er , the best latencies of 5.8 and 5 seconds obtained by NB-IoT and L TE-M respecti v ely using the EDT procedure are mainly due to the multiple xing of the user data with the Message 3 on the dedicat ed traf fic channel, as sho wn in Figure 1 (b) and Figure 2 (b) respecti v ely . NB-IoT and L TE-M towar ds massive MTC: Complete performance e valuation ... (Adil Abou El Hassan) Evaluation Warning : The document was created with Spire.PDF for Python.
316 r ISSN: 2502-4752 4.1.4. Battery life The simulation and system model parameters used to e v aluate the battery life of NB-IoT and L TE-M are gi v en in T able 6 according to [23],[24]. While the assumed traf fic model according to Rel-14 scenario and de vice po wer consumption le v els used are gi v en in T able 7 according to [23], [24]. Also, an Acti v e T imer of 20 seconds is included after connection release where the de vice is in Idle-Light sleep before switching to Idle- Deep sleep, to monitor the do wnlink control channels of NB-IoT and L TE-M i.e. NPDCCH and MPDCCH respecti v ely . Based on the transmission times of the signals and do wnlink and uplink channels gi v en in [23] and using the (3) and (4) with the simulation assumpti on s that are gi v en in T able 7 and a 5Wh battery , the e v aluated battery li v es of NB-IoT to achie v e the MCL of 164 dB in in-band, guard-band and stand-alone operation modes are 11.4, 11.6 and 11.8 years respecti v ely . Whereas the e v aluated battery life of L TE-M to achie v e the MCL of 164 dB is 8.8 years according to the assumed transmission times gi v en in [24]. T o significantly increase the battery life of L TE-M , the uplink throughput should be impro v ed by the increase of the number of base station recei ving antennas, thereby reducing UE transmission time. Therefore based on the simul ation assumptions gi v en in T able 3 where the number of base station recei ving antennas is 4 instead of only 2 according to [21] and the simulation assumptions gi v en in T able 7, the e v aluated battery li v es of L TE-M and NB-IoT are 11.9 and 11.8 years respecti v ely . Kno wing that the 5G mMTC requires battery life be yond 10 years at the MCL of 164 dB, supposing an ener gy storage capacity of 5Wh [7]. Therefore, NB-IoT achie v es the tar geted battery life in all operation modes re g ardless of the antennas configuration of the base station. Ho we v er , L TE-M fulfills the 5G mMTC tar geted battery life e xcept if the number of base station recei ving antennas is 4. T able 6. Simulation and system model parameters for battery life e v aluation P arameter V alue L TE system bandwidth 10 MHz Channel model and Doppler spread Rayleigh f ading ETU - 1 Hz eNB po wer and antennas configuration NB-IoT : 46 dBm (Guard-band, In-band) - 2Tx/2Rx 43 dBm (Stand-alone) - 1Tx/2Rx L TE-M: 46 dBm - 2Tx/2Rx De vice po wer and antennas configuration 23 dBm - 1Tx/1Rx T able 7. T raf fic model and de vice po wer consumption le v els Message format UL report 200 bytes DL Application Ackno wledgment 20 bytes UL report periodicity Once e v ery 24 hours De vice po wer consumption T ransmission and reception po wer consumption P T x : 500 mW - P R x : 80 mW Idle mode po wer consumption P I LS : 3 mW - P I D S : 0.015 mW 4.1.5. Connection density The supported connection density (CDS) that is e v aluated corresponds to the o v erall number of de vices that successfully transmit a payload of 32 bytes accumulated o v er tw o hours with the required latenc y . T o e v aluate CDS of NB-IoT per PRB and square kilometer depicted in Figure 8 (a), the CDS is calculated from (5) using the CRI v alues of Figure 3 and periodicity of connection requests of tw o hours. Re g arding L TE-M, to e v aluate CDS per narro wband and square kilometer sho wn in Figure 8 (b), the CDS is determined from (5) using the CRI v alues of Figure 3, a reporting period of tw o hours and scaling of a f actor 6 corresponding to the L TE-M narro wband (NB) of 6 PRBs. In the tw o scenarios corresponding to the 500 meters ISD sho wn in Figure 8 (a), more than 1.2 million de vices per PRB and square kilometer can be supported by an NB-IoT carrier with a maximum 10 seconds latenc y . Ho we v er , only 94000 and 68000 de vices per PRB and square kilometer can be supported using the (UMA B) and (UMA A) channel models res pecti v ely with an ISD of 1732 meters within the 10-second latenc y limit. Since, in the sc enario of a 1732 meters ISD, the density of base stations is 12 times lo wer than in a 500 meters ISD. Therefore, this dif ference in base station density results in dif ferences of up to 18 times between the connection densities relating to the 500 and 1732 meters ISD scenarios. Indonesian J Elec Eng & Comp Sci, V ol. 23., No. 1, July 2021 : 308 320 Evaluation Warning : The document was created with Spire.PDF for Python.
Indonesian J Elec Eng & Comp Sci ISSN: 2502-4752 r 317 As sho wn in Figure 8 (b), in 500 meters ISD scenario a single narro wband can support up to 5.68 million de vices within the 10-second latenc y limit, by the addition of 2 further PRBs to transmit PUCCH. F or the 1732 meters ISD and (UMA B) scenario, the cell size is 12 times lar ger that e xplains an L TE-M carrier can only support 445 000 de vices within the limit of latenc y of 10 seconds. Also, to further impro v e the L TE-M connection density , the sub-PRB resource allocation in uplink that has been introduced in 3GPP Rel-15 can be used for lo w base station density scenarios. (a) (b) Figure 8. Connection density in relation to latenc y of NB-IoT and L TE-M; (a) NB-IoT and (b) L TE-M 4.2. Comparati v e analysis of NB-IoT and L TE-M perf ormance Figure 9 depicts the diagram comparing the performance of NB-IoT and L TE-M technologies in te rms of co v erage, throughput, latenc y , and battery life that ha v e been e v aluated in Subsec tion (4.1.), on using the same simulation assumptions gi v en in T able 3. Whereas the connection densities of NB-IoT and L TE-M that are compared are the ones e v aluated using the simulation assumptions gi v en in T able 2. The latencies sho wn in Figure 9 are that obtained with the EDT procedure, while the connection densities correspond to the best v alue obtained of the supported intensity of connection requests (CRI) from Figure 3 within the 10-second latenc y limit, and corresponding to the same urban macro scenario using 500 meters ISD and (UMA B) channel model. The 5G mMTC require ment re g arding CRI sho wn in Figure 9 corresponds to the tar geted CRI that is obtained from (5) to achie v e one million de vices per PRB and square kilometer for 500 meters ISD scenario. From T able 4 and T able 5, it can be seen that for both technologies, NPUSCH F1 and PUSCH can be considered as the limiting channels, i.e. the channels that need the maximum transmissi on times to reach the MCL of 164 dB. Indeed, NPDCCH must be configured with 512 repetitions to achie v e the tar geted BLER of 1%, while the maximum configurable repetition number is 2048 in an e xtreme co v erage corresponding to the CE-2 le v el [26]. Whereas, MPDCCH needs to be configured with the maximum configurable repetition number , i.e. 256 re p e titions to reach the tar geted BLER of 1% and the MCL of 164 dB. Therefore, to support operations in e xtreme co v erage, NB-IoT technology can be considered more ef ficient than L TE-M technology . Figure 9. Performance comparison diagram of NB-IoT and L TE-M technologies NB-IoT and L TE-M towar ds massive MTC: Complete performance e valuation ... (Adil Abou El Hassan) Evaluation Warning : The document was created with Spire.PDF for Python.