Grant-less Uplink Transmission for LTE Operated in Unlicensed Spectrum

Grant-less Uplink T ransmission for L TE Operated in Unlic ensed Spectrum Jinyu Zhang 1 , W enting Cha ng 2 , H u aning Niu 2 , Salvatore T ala rico 2 , H o ngwen Y ang 1 1 Beijing Uni versity of Posts and T elecom munication s 2 Intel Mobile Comm unications T e c hnology Ltd. Email: zhangjinyu@bupt.edu.c n Abstract —Deployment of Long T erm Ev olu tion (L TE) in unlicensed spectrum has been a candid ate feature to meet the explosi ve gr owth of traffic dema nd since 3GPP release 13. T o further explore th e advantage of unlicensed bands, in this context the operation of both uplink and downlink has been supported and studied in the sub sequent releases. Howe ver , it has been identified that schedu led uplink transmission perf ormance i n unlicensed spectrum is significantly degraded du e to the double listen-befor e-talk (LBT) requirements at both eNB when send ing the uplin k grant, and at th e scheduled UEs befor e transmission. In this paper , in order to over come th is issue, a n ovel uplink transmission sch eme, which does n ot require any grant, is proposed, and the details regarding the system design are prov ided. B y modeling the dynamics in time of the LBT for both a system t h at employs a conv en tional uplink scheme, as well as th e proposed scheme, it is verified th rough an- alytical evaluation that the double LBT scheme fo r uplin k transmission greatly reduces the channel access probability fo r the UE, and leads consequently to performance loss, while the propose d scheme is able to all eviate this issue. System lev el simulation results, compliant with the L TE standard, show that the proposed scheme can achiev e a significant performance gain in terms of th roughput with negligible perform ance loss fo r the downlink, and oth er technologies operating in the same spectrum. I . I N T R O D U C T I O N Owing the popularity o f smartpho n es, tablets, and other wireless devices, the recent widesp r ead ad option of wireless bro adband has resulted in a treme ndous g rowth in th e volume of mob ile data traffic, which is pr o jected to con tinue unab ated [1], [2]. As a consequ ence of this, the system capacity of w ir eless commun ication systems have been sev erely challeng ed. Howe ver , restricted by the lack of available spectrum resource in licensed ban d, the traditio nal Long T erm Evolution (L TE) techn ology is powerless in tackling this problem . Ther efore, th e av ailab le r esources in unlicensed ban d have attrac ted recently more, and more attention as an importan t com- plement to alleviate the high data traffic load [ 3], [4]. In th is regards, 3GPP h as intro duced its operatio n in unlicensed ban d via Licensed Assist Access (L AA) in release 13 [5], [6]. LAA uses carr ie r aggregation in the downlink to comb ine L TE in unlicensed spectr um with L TE in the licen sed ban d to expand the system bandwidth . While significant ch anges hav e bee n made compare d to the L TE framework through the intro duction of sev eral mechanisms [7], the author s of [ 8] have shown that LAA ensures fair co existence with existing W i-Fi networks. Follo wing the cu rrent mo mentum on unlicensed spec- trum, r ecently 3GPP has started two new working items, named “new rad io (NR) b ased u nlicensed access” [9] and “Enhanc e ments to L TE operation in unlicensed spectrum” [10]. W ith this in m in d, apart from LAA systems, MulteFire (MF) systems [1 1] employ L TE technolog y , but solely work in unlicensed spectrum without assistance o f the “anch or” in licensed spe c tr um. For the se systems the c o ntrol info r mation and refer ence signal must be supp orted to be transmitted on un licensed carriers along with the entire d a ta. In this regards, a MF system is totally different than an LAA system , and its system framework needs to be modified to suppo rt the sole ope r ation in u nlicensed b and. Even tho ugh the MF tech nology is still at an embryo nic stage, th e combinatio n of L TE like perf ormance benefits, a n d W iFi like deploymen t simplicity makes MF a significantly importan t sup plement, and valuable study topic to meet the ever -increasing wireless tr affic. In legacy L TE sy stems, scheduled based up link (SUL) transmission has been considere d in the 3GPP release 14 study , wherein uplin k tran smission is conditional to an explicit uplink grant via p hysical d ownlink con trol channel (PDCCH) [12]. In ord er to comply with the FCC regulation requ irements, an d in order to maintain fair coexistence with oth er technolo gies, th e listen-b e fore- talk (LBT) mechanism is applied to check whether the channel is clear or o ccupied before usin g it. Howe ver, the use of the legacy two stage mod ality for LBT redu ces the uplink chan nel access pr obability . This drawback is highligh ted an d verified in terms of channel access probab ility by mo deling the dynamics in time o f a system employing LBT throu ghout a Markov chain, similarly as [13]– [ 15]. Besides th e penalty impu ta b le to a lowered ch annel access probab ility , the perfor mance of SUL is also negatively affect by the p rocessing delay (genera lly 4 ms d ue to hardware constraints) between the uplink grant and the scheduled tran smission, wh ich m ay also lead to tr a nsmission latency and reso urce waste in case there is no d ownlink data. H e nce, scheduled b ased 11 10 6 9 8 7 1 2 3 4 5 12 13 empty First Partial SF MCOT Fig. 1. Schedu led based uplink transmission mode uplink tran sm issions are not suitable in un licensed band. In th is paper, we con sider a new up link transmis- sion scheme, which does no t require any eNB sch edul- ing grant, named grant-less up link (GUL) transmission. While this methodo lo gy highly resemb les that curre n tly used in th e W iFi uplink d e sig n, it is a significant d epar- ture fro m the existing SUL transmission of the legacy L TE. In this regards, a n umber of e n hancemen ts, which are discussed alon g this manuscr ipt, need to be ma d e with r espect to the legacy L TE design in order to b e able to prop erly ena b le and perform GUL transmissions. The rest of this paper is organize d as follows. Sec- tion II begins with a brief introductio n of the SUL scheme. Th is section continues b y building an analytical framework based on a Ma r kov chain mode l, which is employed to model the dyn amics in time of the LBT proced u re fo r both the user equip ment (UE) and eN B. This analy tical fr amew ork is then used to com pare th e channel access p robability of SUL and GUL schemes, and high light the benefits of th e propo sed scheme. T he overview p rocedu r e an d d esign d etails fo r GUL mod e are then provided in section III. In section IV, the perfor mance of the pro posed sche me is ev aluated via system level simulatio ns. Finally , con clusions are drawn in sectio n V. I I . S Y S T E M M O D E L In legacy L TE, the UE that intends to tran smit data needs to o btain an uplink gra nt f rom th e serving eNB, and on ly then it can start u plink transmission, as il- lustrated in Fig.1. In primis, the eNB is r equired to perfor m Cat.4 LBT on the target carr ier for the u plink grant transmission , as regulated in [ 1 7], [ 18]. Once it is able to access successfully the chan nel, th e sub sequent maximum channel occupan cy time (MCO T) can b e occupied , and sched uled for either downlink or u plink transmission by the eNB. While the PDCCH carrying uplink gran t can be transmitted in the fir st a vailable subframe (SF), due to the processing delay th e physical uplink shared cha n nel (PUSCH) is scheduled at the latter SFs du ring the same MCOT . The rem a in ing symbols in the downlink SFs can be utilized for downlink data transmission, if any . Before u plink transmissions can take place, the schedu led UE needs to complete an additional LBT (either sin gle inter val LBT or Cat.4 LBT) [18] afte r receiving the gr ant. If this second LBT fails, the resou rces reserved fo r uplin k are wasted. Intuitively , the SUL mode ham pers the channel access probab ility f or the UE. In order to overcome this issue, it is proposed h ere to ad opt on e-LBT u p link acc ess mechanism instead of this d o uble-LBT proced ure, which leads to up link transmissions tha t ca n be pe rformed autonom ously with o ut req uiring grants, which we ref er to as G U L . For the propo sed GUL sch eme, on the o ther hand, similar ly to SUL, Cat.4 LBT is still em ployed for the fair sharing of unlicensed band. While M arkov chains and their prop erties have b een extensi vely used to model and ch aracterize the pro ce- dure of LBT for W i-Fi and LAA [13]–[16], in this contribution they are used to mod e l the LBT fo r MF systems, in orde r to study its coexistence with W i-Fi. In p articular, utilizing a M arkov cha in mod el, the LBT proced u re of a WiFi, and MF acce ss node is mo deled, and th e transmit prob a bilities in a ran domly ch osen slot time can be calculated by (1) and (2) , r espectiv ely [14], [15]. For the se equatio ns, Q = 2(1 − p b )(1 − p f )(1 − 2 p f ) , p W iF i tx = 2 q (1 − p b )(1 − 2 p f ) 2(1 − p b )(1 − p f )(1 − 2 p f ) + q [ W 0 p f (1 − (2 p f ) m ) + (1 + W 0 − 2 p b )(1 − 2 p f )] (1) p C at 4 tx = 2 q (1 − p b )(1 − p f ) R Q + q [ W 0 P (1 − p f )(1 − (2 p f ) ( m + 1) ) + P R (1 − 2 p b )(1 − 2 p f ) + 2 R (1 − p b ) 2 (1 − p f )(1 − 2 p f )] (2) P = ( p b + p f − p b p f ) , R = (1 − p ( m +1) f ) , q d enotes the pr obability of packet ar riv al, m is the maximum clear channel assessment (CCA) stage, an d W 0 is the initial contentio n window size. p f and p b denote the probab ility o f transmission failure due to collisions, and the pr o bability that the channel is detected to be occupied , respectively . In [14], [15], these proba b ilities are although determined und er the simplified assumption that all th e node s in the coexistence scenario can detect the signal fr om all othe r nod es ab ove the carr ier sense threshold. In or der to ad dress this issue, the d istribution of the detected energy [ 1 9]–[21] is here taken into account. For simplicity , let assume th at the path loss between any two nod es are identical. The total receiving power P r x can be then be o btained by multiply ing the receiving power P 0 r x from a sing le transmitting node by th e total numb er of the transmitters n . T hus, th e distribution of the energy d etection co nditioned o n the number o f transm itter s could be expressed as follows f Y ( y | n ) = ( 1 2 µ Γ ( µ ) y µ − 1 e − y 2 idle 1 2 ( y 2 γ ) µ − 1 2 e − 2 γ + y 2 I µ − 1 ( √ 2 γ y ) busy (3) where γ is the sign al-to-no ise ratio (SNR), and de- pends o n the number of transmitters n since γ = nP 0 r x /P noise , Γ ( . ) repre sen ts th e gamm a function , an d I v ( . ) is the v th-order modified Bessel f unction of the first kin d. Assume that the chann el sense failure an d the tra n s- mission collision both occu r in the case th at the detected energy is ab ove th e L BT threshold y thv . In a n e twork with N ac c ess n odes, it yields th at p b = p f = N − 1 X n =1  N n  p n tx (1 − p tx ) N − 1 − n Z + ∞ y thv f Y ( y | n ) dy . ( 4) The transmission prob ability for a W iFi Access Point (AP), and a system with Cat.4 LBT is evaluated then by solvin g (1) and (2) using (4 ) , respectively . For a SUL scheme, on ly N eNBs p erform Cat.4 LBT , an d the u plink channel is av ailab le only when bo th downlink Cat.4 LBT , and the single slot LBT a t the UE side succ e e d. Th u s, in this case the u plink channel access pr obability can be expressed as p S U L tx = (1 − p b ) p C at 4 tx (5) In the proposed GUL mode , the UE perform s inde - penden t Cat.4 L BT , which is n early th e sam e b ehaviour as eNB in respect to the chan nel access p rocedu re. Therefo re, the channel access probab ility for both UE and eNB can be obtained b y sub stituting within N in (4) th e sum of the num ber o f UEs and eNBs inv olved. Fig. 2 shows th e chann el ac c ess pro bability b ased on the assump tio ns th at y thv = − 72 dBm for bo th the W i- Fi an d the MF system. T h e nu m ber of WiFi APs or M F eNBs dep loyed for each ope rator is N = 5 , and each eNB only has o ne a c ti ve UE . For this plot, m = 4 , 3 U R E D E L O L W \ R I S D F N H W D U U L Y D O & K D Q Q H O D F F H V V S U R E D E L O L W \ : L ) LL Q : L ) L 6 8 / í 0 ) 0 ) 8 / L Q : L ) L 6 8 / í 0 ) : L ) LL Q : L ) L * 8 / í 0 ) 0 ) L Q : L ) L * 8 / í 0 ) Fig. 2. Channel access probability W 0 = 16 , P 0 r x /P noise ≈ 10 and the f o llowing two scenarios ar e shown: • W iFi+SUL-MF: W iFi APs o f operator 1 co exist with MF eNBs and UEs of ope rator 2, wh ich operates in scheduled b ased up lin k modality; • W iFi+GUL-MF: W iFi APs of op e rator 1 coexist with MF eNBs and UEs of ope rator 2, wh ich operates in grant- less based uplin k modality . As illustra te d by Fig. 2, the SUL sch eme is subject to a small uplink ch a nnel acc ess pr o bability du e to the doub le LBT requ ir ed, while fo r th e GUL mo d e this imp roves significantly with negligible impac t on the W iFi system perfor mance. I I I . S Y S T E M D E S I G N O F G R A N T - L E S S U P L I N K M O D E Fig. 3 provides a n illustration of the overall pro cedure for the GUL mode. Fir stly , the UE with u plink data perfor ms cha n nel sen sing. In this ca se, Cat.4 LBT is adopted to maintain the fair coexistence with the incum - bent sy stem and other techno logies. A pr eamble sign al is needed befor e d ata tra nsmission for th e d e tection at the anchor e d eNB, an d signalling of con trol inform ation. A r e ser vation signal may a lso b e needed to align with the pr edefined boun dary . Th en, the UE c a n use the whole MCOT fo r data transmission rath er than shar e d with downlink. Finally , the eNB need s to fe e d back the A CK/N AC K inf ormation for Hybrid Autom a tic Repe a t Request (HARQ) proce ss. The pr oper use of GUL mode nee d s a quite different framework than that used b y SUL in the L TE technol- ogy . Therefor e, a nu mber o f en hancemen ts, such as the control inform ation an d f eedback, are neede d with the respect to the legacy L TE design , an d the details ar e discussed in this section. A. Detectio n of PUSCH at eNB side Due to the lack of scheduling , the serving eNB is not aware of the UE’ s transmission and it nee d s to detect the p resence o f the uplink burst. T wo can didate me th ods can be taken in to conside ration f or such indication: Part ial Super S Asynchronou UL MCOT Synchronous Fig. 3. Grant-l ess uplink transmission m ode • Implicit ind ication b y demodu lation refere nce sig- nal (DMRS): the serving eNB per forms b lind de- tection of the DMRS seq uence to infer th e presence of PUSCH; • Explicit in d ication throug h Uplink Contro l Ind icator (UCI): in this c ontext, the existing UCI for mats can be reused to provide additio nal info rmation regarding th e uplink b urst. The con tent of UCI includes but is not limit to the following field s: HARQ pr ocess nu mber, UE id e n tifier , an d new data indicator (NDI ). B. Up link Sub -frame Design Since the LBT could b e comp leted at any time instant, mostly not alig n ed with the primary cell (PCell) SF bound ary , th is m ay result in a waste of resourc e s due to the fact that the transmission is p ostponed until the bound ary of next SF . I n order to better u tilize the in te r val of time from the end s of th e L BT until the PCell sub- frame boun dary , a mor e flexible desig n of the uplink SF is req uired. As shown in Fig. 3, three up link SF ty pes can be ado p ted: • Synchr onous SF , which is alig ned with the boun d- ary of PCell SF to minim ize the implem entation impact. I n this context, a partial SF o r super SF can be defin e d on a subset of O FDM sy mbols within th e uplin k SF ( similar to the partial T TI for downlink LAA ) , while the PCell still remains aligned in terms of tim ing relationsh ip with the uplink burst tran smission. In this case, the UE can start PUSCH tran sm ission at certain known OFDM symbol positions within a SF with th e aim to limit the UE schedu ling co mplexity . In particular, as UE may know the duration of the partial TTI, it may need to create multiple poten tial par tial SFs correspo n ding to different hypo thesis o f possible partial SF . Howe ver, this incur s in a significant computatio n and b uffer co mplexity at the UE side. Thus it is desirab le to limit the set of possible starting positions to a ssum e some p redefined and restrict values, e.g. { 1, 8 } . • Asynchr onous SF , which cann ot be align ed with PCell bo undary as illustrated in Fig. 3. As long as the c h annel is acq uired th rough LBT , UE could carry out the u plink tr ansmission based on the legacy 1 ms SF design. C. Schedu ling, Link Adapta tion an d HARQ Operation Instead of r elying on th e indication from the serv ing eNB, th e UE n eeds to autono mously select the r esource allocation in GUL mo d e. Accurate chan nel state in f or- mation (CSI) is essential for both the sch e duling at the UE side, and the d emodula tio n at the eNB side. Ap art from this, th e UE also needs feedb ack inform ation f o r HARQ retransmission. In this regards, the proc e ss could be summ arized in following step s: • Step 1: eNB estimates and calculates the up link CSI based o n the soundin g ref e r ence sign als (SRSs) from the UE. I n particular, in th is case the legacy L TE design for SRS can b e reu sed, and they can b e transmitted in the last OFDM symbol. Ad ditionally , the CSI r equest can be tra n smitted alo n g with SRSs. • Step 2 : The UE choo ses an a ppropr iate mo dulation and codin g scheme (MCS). The selection can b e done by the eNB, wh ic h can ind icate th e b est suitable MCS to UE. Alternatively , the UE can request CSI, and based upon this info r mation it can select the approp riate MCS by itself. • Step 3: The UE tran smits data alon g with th e scheduling inform a tion via PUCCH, which may contain the HARQ process num ber and NDI. • Step 4: Th e eNB transmits AC K/N ACK feed back via PDCCH, after receiving up link data. For link ad aptation, the possible suitable optio ns a r e: • The eNB dy n amically feed backs the up link CSI for MCS selection, wh ile indicatin g HARQ A CK/N AC K fe e d back; • The UE uses the MCS indicated in latest DCI. D. Contr ol Cha nnel Design When UEs have simultaneou s up lin k d ata an d contro l transmission, contro l signa lin g c an be multiplexed with Fig. 4. Illust ration of PUCCH control regi on data prio r to the discrete fourier transfo rmation (DFT) to preser ve the single- carrier proper ty in the uplink transmission as shown in Fig.4. This metho dology c an be reu sed in systems such as MF , wh ich work solely in unlicensed spectrum, but with cer tain extent. In fact, in these systems a different content can b e carried , and a list of possible fields is as follows: • Cell r adio ne twork tem porary id entifier (C-RNTI); • HARQ p rocess nu mber; • NDI, which is used to state wheth er the curren t transmission is a retransmission or not; • MCO T and uplink burst related inform ation d e- scribed as the numb er of SFs. I n case not all the SFs are used, the remainin g SFs co u ld be sche d uled by eNB for downlink or uplin k tran sm ission o f other UEs. • Carrier used; • A-CSI, an d HARQ A CK/N ACK bitmaps. In order to reduc e the UCI signallin g overhead, it is p r eferred to transmit som e of this in f ormation once, especially for fields such as A-CSI an d HARQ A CK/N AC K bitmaps, while some othe r fields which are essential (i.e, H A RQ process numbe r, C-RNTI and NDI) should be tra nsmitted in each SF . In th is regard , at least two d ifferent sizes for UCI should be p r edefined: o ne which includes the complete UCI with all field s, and one which incorpor ates o nly necessary field s. Since bo th the MCS in dex (which can b e determ ined accordin g to subsection III - C) and th e UCI size ar e n eeded to separate the control in formation , th e eNB can perfo rm blind detection to determin e this last information . Th is op tion is although very com putational intense, an d alternatively the HARQ ACK/N A CK or the rank ind icator (RI) can be used to indicate the UCI size. I V . S I M U L A T I O N A N D P E R F O R M A N C E E V A L U A T I O N This section provid es the results obtain ed from com- prehen sive system level simu lations per formed with the − 1000 − 800 − 600 − 400 − 200 0 200 400 600 800 1000 − 800 − 600 − 400 − 200 0 200 400 600 Macro eNB small cell eNB Fig. 5. Network layout for the indoor scenario T ABLE I M A I N S I M U L AT I O N P A R A M E T E R S Parame ters V alue Scenari o Layout Indoor Scenario Number of UE 20 UEs per sector Channel Model WINNER B+ Carrier frequency 5GHz Inter -Stati on Distance 500m MCO T 5ms Tra f fic Model FTP Model 3 File Size 0.5MByte DL:UL Traf fic Ratio 50:50 eNB Tx Power 18dBm eNB Antenna Gain 5dB UE T x Power 18dBm UE A ntenna Gain 0dB aim to ev alu ate th e perfo rmance o f the p roposed scheme. The simulations are perfor med under the assumptions agreed in [6], wh ich are summarized in T able I. An indoor d eployment with 7 hexag onal cell sites is con- sidered for each operato r , as shown in Fig.5 . Every site h as 3 sectors with 4 MF eNBs or 4 Wi-Fi APs random ly dropp ed, an d gro uped as a cluster . Similarly to Fig. 2 two scenario s are con sidered: W iFi+SUL-MF , and W iFi+GUL- MF . Fig. 6 shows the perform a nce in terms of the mean user p e rceiv ed thro ughpu t ( U PT ) for these two scenar io s fo r b oth a Wi Fi and MF systems and for both uplin k and downlink. This figure highligh ts that the uplink thro ughpu t of a MF o perator is quite low when the SUL scheme is adopted f o r th e aforementio ned issues, a n d th at the propo sed GUL scheme allows to improve sign ifica n tly its perfor mance u p to a c hieving co mparable pe r forman c es with W iFi. As for the d ownlink, th e prop osed schem e leads to a slight per formanc e lo ss, which is n egligible compare d to the gain obtained for u p link. I n fact, per for- mance in terms of sum thro u ghput of both downlink and uplink are still significantly imp roved with the p roposed scheme. Moreover , by comp aring the p erforman ces of W iFi in terms of th rough put between the case wh en 0 5 10 15 20 25 30 35 40 45 1 1.15 1.3 1.6 1.75 1.9 2.05 2.5 UPT (Mbps) Total Offered Load for each UE (Mbps) WiFi in WiFi+SUL-MF WiFi in WiFi+GUL-MF MF in WiFi+SUL-MF MF in WiFi+GUL-MF (a) A vera ge uplink UPT 0 20 40 60 80 100 120 1 1.15 1.3 1.6 1.75 1.9 2.05 2.5 UPT (Mbps) Total Offered Load for each UE (Mbps) WiFi in WiFi+SUL-MF WiFi in WiFi+GUL-MF MF in WiFi+SUL-MF MF in WiFi+GUL-MF (b) A verage do wnlink UPT Fig. 6. A verage UPT performance the SUL and the GUL schem e is used, it is possible to notice that the pro posed sche me is able to still guar antee coexistence between MF and W iFi. For both downlink and up link, the WiFi p e rforman ce is sligh tly degraded due to the inten se channe l acce ss competitio n with MF . Howe ver, such p erforma n ce d egradation is accep tab le, and it may also incur in case the numb er of W iFi APs are increased in a cer tain a rea. In con clusion, th e propo sed GUL scheme c a n achieve rem arkable per formanc e g ain for MF u plink and m aintain th e f riendly co existence with incumben t M F downlink, and Wi-Fi technolo gy . V . C O N C L U S I O N In this pape r, in o rder to cope with the tremend ous deterioratio n of th e u plink perf ormance of L TE systems operating in unlicensed spectrum , suc h as MF , a n ew transmission sch eme is prop o sed, whic h allows to per- form gr a nt-less tr a n smissions. By d ev eloping an an alyt- ical framework ba sed on a Mar kov chain representation of th e LBT proc edure, it shows that the GUL schem e is able to increase the u plink channel acc e ss pro bability in a MF system co mpared to a sch e duled based scheme. In addition, along the p a per system designs and details on how to enable this transm ission sch eme within the L TE ecosystem are e la b orated. Fina lly , compre h ensiv e system lev el simulations are provided, an d ev aluation indicates that the pro posed GUL mode can lea d to a sign ificant improvement of the up link UPT p erforman ce with th e negligible per forman c e loss for MF downlink and W i-Fi systems. REFERENCES [1] Cisco System, Cisco V isual Networking Index: GlobalMo bile Data Traf fic Forecast Update, 2015-2020, [Online]. A vai lable : https:/ /www .cisco.com/c/e n/us/solutions/collateral/service- provi der/visua l-net working-inde x-vni/mobile-white-paper-c11- 520862.pdf [2] Nokia Whitepaper: “Enhance mobile networ ks to deli ver 1000 times more capacity by 2020”. [3] Huawei Whitepa per: “U-L TE : Unlicensed Spectrum Utiliz ation of L T E”. [4] Qualcomm W hitepa per: “Extending the Benefits of L TE -A to Unlicen sed Spectrum”. [5] 3GPP TD RP-141664,“Study on Licensed -Assisted Access using L T E, ” Sept. 2014. [6] 3GPP TR 36.889 3rd Generation Partnershi p Project; T echni cal Specifica tion Group Radio Access Network; Study on Licensed- Assisted Access to Unlicensed S pectrum;(Relea se 13)[R]. 3GPP V13.0.0 (2015-06). [7] A. Al-Dulaimi, S. Al-Rubaye, Q. Ni and E. Sousa,“5G Com- municati ons Race: Pursuit of More Capa city Trigge rs L TE in Unlicen sed Band, ” in IEEE V ehicular T echnology Magazi ne, vol. 10, no. 1, pp. 43-51, March 2015. [8] H. J. Kwon et al.,“Licensed-Assisted Access to Unlicense d Spec- trum in L TE Release 13, ” in IEEE Communications Magazine , vol. 55, no. 2, pp. 201-207, F ebruary 2017. [9] 3GPP TD RP-170828, “New SID on NR-based Access to Unli- censed Spectrum, ” March 2017. [10] 3GPP TD RP-170848, “New W ork Item on E nhance ments to L TE operati on in unlicensed spectrum, ” March 2017. [11] https://www .multefire.or g/ [12] R. K araki et al.,“Uplink Performance of Enhanced Licensed Assisted Access (eLAA) in Unlicense d Spectrum, ” 2017 IE EE W irele ss Communicat ions and Networ king Confere nce (WCNC), San Francisco, CA, March 2017. [13] G. Bianchi,“Performan ce analysis of the IEEE 802.11 distrib uted coordina tion function, ” in IEEE Journal on Selecte d A reas in Communicat ions, vol. 18, no. 3, pp. 535-547, March 2000. [14] C. Chen, R. Ratasuk and A. Ghosh,“Do wnlink Performance Analysis of L TE and W iFi Coexist ence in Unlicensed Bands with a Simple Listen-Be fore-T alk Scheme, ” IEEE V ehicul ar T echnology Conferen ce (VTC Spring), Glasgow , UK, May 2015. [15] Y . Gao, X. Chu and J. Zhang,“Performance Analysis of LAA and W iFi Coexist ence in Unlicensed Spectrum Based on Marko v Chain, ” IEE E Global Communicati ons Conference (GLOBE- COM), W ashington, DC, D ec. 2016. [16] Z. Fu, W . Xu, Z. Feng, X. Lin and J. Lin, ”Throughput Anal- ysis of L TE-Licensed-Assisted Access Networks with Im perfec t Spectrum Sensing, ” 2017 IEEE W ireless Communicatio ns and Networ king Conference (W CNC), San Francisco, CA, March 2017. [17] ET SI EN 301 893 Broadband Radio Access Networks (BRAN); 5 GHz high performanc e RLAN; Harmonize d EN cov ering the essentia l requireme nts of article 3.2 of the R&TTE Directi ve [S]. ETSI V1.8.1 (2015-03). [18] 3GPP TR 36.213 3rd Generation Partnership Project; T echnical Specifica tion Group Radio Access Network; Evolv ed Uni ve rsal T errestrial Radio Access (E-UTRA); Physical layer procedu res (Relea se 14). 3GPP V14.2.0 (2017-03). [19] F . F . Digham, M. S. Alouini and M. K. Simon,“On the Energy Detect ion of Unkno wn S ignals Over Fadi ng Channels, ” in IEEE Tra nsactio ns on Communications, vol. 55, no. 1, pp. 21-24, Jan. 2007. [20] F . F . D igham, M. S. Alouini and M. K. Simon, “On the energy detec tion of unknown signals over fading channels, ” , IEEE Interna tional Conference on Communications (ICC), Anchorage, AK, May 2003. [21] B. Li, T . Zhang and Z. Zeng, “LBT with adapti ve threshold for coe xistenc e of cellular and WLAN in unlice nsed spectrum, ” IEEE Interna tional Confere nce on Wire less Communications & Signal Processing (WCSP), Y angzhou, China, Oct. 2016.