Research Article. ISSN (Print) O.S.Popov, Ukraine. *Corresponding author Ali Al-Anssari

Transcription

1 Scholrs Journl o Engineering nd Technology (SJET) Sch. J. Eng. Tech., 2014; 2(5B): Scholrs Acdemic nd Scientiic Publisher (An Interntionl Publisher or Acdemic nd Scientiic Resources) ISSN X (Online) ISSN (Print) Reserch Article Multi-Period LTE RAN Plnning with Opertor Proit Mximition (Prt 1: Optimition Model Synthesis) Dmytro Ageye 1, Ali Al-Anssri* 2 1 Telecommuniction systems deprtment, 61166, hrko Ntionl Uniersity o Rdioelectronics, Ukrine. 2 Telecommuniction systems deprtment, 65029, Odess Ntionl Acdemy o Telecommunictions nmed ter O.S.Popo, Ukrine. *Corresponding uthor Ali Al-Anssri Emil: Abstrct: The LTE RAN plnning is ery itl step or wireless communiction technology. The cretion o network is not n instntneous process; it typiclly goes stge by stge. In this rticle (irst prt) n optimition model or the LTE RAN network plnning problem is described. Рrmeters such s requirements, the cost o instlltion nd usge o equipment, the speciic income rom proiding serices my chnge oer time in the gien optimition model. The nlysis o experiment's results showed tht ormultion o the problem, which is shown boe nd presented s problem o MILP, llows to obtin the correct solutions rom the prcticl point o iew. Comprison o the opertor's proit during the single-period nd multi-period plnning showed tht usge o multi-period plnning increse the proit mrgin to 12% more thn the single-period plnning. eywords: LTE RAN, multi-period plnning, test-point, Signl-Intererence-Noise Rte, optimition model INTRODUCTION The Long Term Eolution(LTE), lso known s Eoled Uniersl Terrestril Rdio Access (E-UTRA), is step towrd the4thgenertion(4g) o mobile rdio technologies to increse the spectrl eiciency nd to obtin higher throughput. The min dntges with LTE re high throughput, low ltency, plug nd ply, FDD nd TDD in the sme pltorm, n improed end-user experience nd simple rchitecture resulting in low operting costs. The combintion o powerul technologies like OFDM nd MIMO techniques in the sme system increses spectrl eiciency, nd improes link relibility without dditionl bnd width or trnsmit power[1]. Rdio network plnning is ery itl step or wireless communiction technology. As stndrdition work o LTE is pproching the end line, it s high time to go or eicient rdio network plnning guideline or LTE. Designing, deploying, nd mnging ny cellulr system requires cler objecties to be identiied rom the outset. These include deinition o the serice re; the projected number o customers; their distribution, spectrum ilbility, growth rte, nd system usge; nd the network interconnect greement, numbering, nd routing policy or inter network ccess nd roming. In [2] coerge nd cpcity estimtion is crried out in rdio network dimensioning. Rdio link budget is inestigted or coerge plnning. Theoreticl work is lter put into the deelopment o n Excel bsed dimensioning tool which is designed to keep the interce simple nd to set the unctionl prts clerly distinguishble. It should be noted tht the cretion o network is not instntneous process. Typiclly, the cretion o network goes stge by stge, with grdul increse in its cpcity nd territory coered. This mkes the problem o multi-stge plnning ensuring high economic eiciency urgent. As criterion o eiciency in this cse we cn select the mximum o opertor's proit. Multi-period design [3] reers to network design problems tht spn oer time horion in terms o weeks to months, nd sometimes een to seerl yers. When we consider multiple periods, there re ew importnt issues to consider (besides the incrementl demnd olume): 774

2 Dmytro Ageye et l., Sch. J. Eng. Tech., 2014; 2(5B): ) The cost structure my chnge oer time (e.g., due to economic discounting); 2) Demnd routed in one time window my he mintennce cost in subsequent time window; 3) The cpcity expnsion cn hppen oer the entire time horion. In [4] proposed method o WIMAX network plnning with opertor's proit mximition, which cn be used s bsis or the deelopment o our method multi-period LTE RAN plnning with opertor proit mximition. Our work will be diided into seerl prts. The irst rticle ocuses on the optimition model synthesis o multiperiod LTE RAN plnning with opertor proit mximition. The second rticle will nlye the eectieness o the proposed method nd peculirities o its ppliction will be identiied. In this rticle (irst prt), we describe n optimition model or the LTE RAN network plnning problem. To do this, the physicl nd rdio-electricl prmeters relent to the model re irst identiied nd described. Such prmeters re then ssocited with binry nd semicontinuous decision ribles. Logicl reltions, coerge, nd cpcity requirements re represented by liner inequlities in the decision ribles. Рrmeters such s demnds, the cost o instlltion nd usge o equipment, the speciic income rom proiding serices my chnge oer time in our optimition model. MODEL FOR LTE NETWOR PLANNING A stndrd model, suitble or plnning purposes, identiies wireless network with set o trnsmitting nd receiing ntenns scttered oer territory. The network design process consists in estblishing loctions nd suitble rdio-electricl prmeters o the ntenns. The network plnning process requires n dequte representtion o the territory. Such model cn be represented s set o geometricl entities, nmely test points (TP). In the TP model, grid o pproximtely squred cells is oerlpped to the trget re. Antenns re supposed to be locted in the center o test points: ll inormtion bout customers nd QoS in TP, such s tric demnd nd receied signls qulity, re ggregted into single coeicients. The TP model llows or smrter representtions o the territory, o the ctul ntenns position, o the signl strengths, nd o the demnd distributions. For estimtion o signl qulity nd chnnel throughput cn be used Signl-Intererence- Noise Rte (SINR). Required SINR is the min perormnce indictor or LTE RAN. Cell edge is deined ccording to the Required SINR or gien cell throughput. Required SINR depends up on the ollowing ctors: Modultion nd Coding Schemes (MCS); Propgtion Chnnel Model. 1 SINR LinkLost( own) SINR RequiredSINR - kown 1 LinkLost( k) CellBW MxTxPwr ThermNoiseDens RxNoiseFig 1 (1) (2) Using Required SINR indictor we mke cpcity plnning. OPTIMIZATION MODEL SYNTHESIS AND PROBLEM DEFINITION LTE RAN cn be represented s set o enodeb which trnsmit inormtion to UE. For plnning gol we modeling UE s TP s described boe. Denote it: A - set o test points, corresponding to set o equl-sied squre res coering n re o the LTE RAN. Ech enodeb typiclly consisting o pylon ccommodting number o trnsceiers (TRX). Eery TRX is chrcteried by set o rdio-electricl prmeters s requency chnnel, which belongs to inite set o ilble chnnels, ech hing constnt bndwidth; emitted power t which TRX trnsmits on gien requency. Process o creting LTE RAN is multi period, with known number o stges. Denode: Z -s et o trnsmitters tht cn be mounted on the enodeb`s in loctions cndidtes; requency chnnel is used to orgnie the rdio link between the UE nd enodeb to the LTE RAN. 775

3 Dmytro Ageye et l., Sch. J. Eng. Tech., 2014; 2(5B): F inite set o ilble chnnels; W constnt bndwidth o requency chnnels; P emitted power t which TRX Z trnsmits on gien requency F. In eery site nd or eery requency chnnel, it is possible to instll either single TRX, mounting n omni directionl ntenn, or seerl TRXs, ech with directie ntenn. Note tht the ctition o TRX my preent the simultneous ctition o other potentil TRXs in the sme site nd operting on the sme requency. This is the cse, or exmple, o TRXs with similr imuths. Deine G 1, G 2,, G - mily o sets, where G i Z, i 1, Cretion o LTE RAN is multistep process with limited budget. We write number o periods Q Determintion o the required leel o costs c represent the oerll cost o instlltion o TRX oer period k. c, is set o mutully exclusie TRXs. - represent the usge cost o TRX, which ctited on requency oer period k. Demnds rom subscribers deined or eery test point A nd chnges during cretion o LTE RAN: h - demnd rom subscribers o test point A during period k. In proiding serices opertor receies reenue per unit olume A cn be deined s: e Ek h E k. In this cse reenie rom sered TP. (3) LTE supports number o combintions o modultion schemes nd orwrd error correction (FEC) coding schemes [5]. These combintions re reerred to s burst proiles. By deining dierent burst proiles, dierent spectrl eiciencies cn be reched [6]. For eery proile V two prmeters, re lso introduced, where representing the SIR threshold tht must be reched to ensure serice coerge; nd is the spectrl eiciency [ bit ] ssocited with the burst proile. sh We introduce sets o Boolen ribles: 1 i ТТ A is sered t plnning period k, y 0 otherwise. (4) 1 i ТRX Z instlled on plnning period k, x - 0 otherwise. (5) 1 i TRX Z use requence chnnel t plnning period k, y 0 otherwise. (6) 1i ТТ A is sered by TRX Z on requence F x with burst proile V t plnning period k, 0 otherwise. (7) We introduce lso set o ribles: p representing the power emitted by TRX Z on chnnel F t plnning period k. TRX Z cn be ctited o ny requency only i it hs instlled beore. This cn be expressed by the ollowing liner constrints: 776

4 Dmytro Ageye et l., Sch. J. Eng. Tech., 2014; 2(5B): whereм lrge constnt. F k1 y M k1 x, Z, (8) TRX Z A test point cn be instlled only one times: k1 x 1, Z, (9) A cn be sered only i there exists t lest one TRX sering on requency with burst x, F, V. This cn be expressed by the ollowing proile, tht is t lest one rible 1, or Z liner constrints: y x( k), k 1... (10) Z F V I x 1, or some Z, F, V, then TRX must be ctited on requency : x y Z, F, V, k 1. (11) The plnning process requires the doption o propgtion chnnel model (1), (2). Obsere tht i x 1, or some A, Z, F, V, then is sered by β on requency with proile nd the corresponding SIR inequlity (1), (2) must be stisied. We cn esily rewrite in the ollowing liner orm: p p M 1 x ( k, (12) where N ) Z \{ } - oerll strength ttenution [ 0, 1] A ; N - is the therml noise; М lrge constnt. to the center o ech TP rom the center o the test point ccommodting trnsmitter Z When TPаsered by LTE RAN, tht is to sy x 1, then tric occupies prt o bndwidth Wo chnnel. The totl bndwidth required to serice consumed tric must not exceed the bndwidth o communiction chnnel: 1 h x W Z, F, k 1. (13) AV To preent the ctition o mutully exclusie TRXs, we introduce the ollowing mily o constrints: y 1, Z, F, k 1. (14) G In cse i TRX is not ctited on requency then p 0. This cn be expressed by p mx y P, Z, F. (15) We need to dd to optimition model tht describing the problem, the constrint o the budget lloction between periods. Expenses t ech stge consists o the cost o instlling new TRX: c x nd usge cost o Z c y. Thus or the totl budget ) TRX, which ctited on requency : Z q (k or the k-th stge cn be written: q c x c y. (16) Z F Z The totl cost o creting the entire network LTE RAN must not exceed the mximum lue o Q: k1 q Q. (17) 777

5 Dmytro Ageye et l., Sch. J. Eng. Tech., 2014; 2(5B): In the process o plnning LTE RAN is necessry to ind network conigurtion to ensure mximum opertor proit. Tht cn be represented s ollows k1 A E k h c x c y mx k1 Z k1 Z F We re inlly ble to summrie the oerll MILP ormultion k1 A E k h c x c y mx k1 Z k1 Z F (18), (19) s.t. p p M 1 x N, A, Z, F, V (20) p Z \{ } AV x y x( k), k 1.. Z F V 1 h x G mx y P y W (21) Z, F, k 1 ; (22) 1, Z, F, k 1 ; (23) y Z, F, V, k 1 (24) F k1 y k1 M k1 x 1, x, Z Z, Z, F, k 1, (27) (25) (26) [ c x c y ] Q (28) k1 Z F y 0,1, A, k 1 x 0,1, Z, k 1 y 0,1 (29) (30), Z, F, k 1 (31) x 0,1, Z, F, V, k 1 (32) To sole the problem (19)-(32) o mixed integer progrmming oered to use mthemticl modeling nd clcultion sotwre, such s ILOG CPLEX A BRIEF ANALYSIS OF THE PROPOSED OPTIMIZATION MODEL The nlysis o experiment s results showed tht ormultion o the problem, which is shown boe nd presented s problem o MILP, llows to obtin the correct solutions rom the prcticl point o iew. We perormed series o experiments or comprison o optimition models o single-period [4] nd multiperiod plnning, which is oered by us. The experiment ws perormed t the ollowing lues o input dt: -lues o time independent prmeters were the sme; -lues o time-dependent prmeters ( c, c, h ) or single-period plnning were equl to the lues or the lst period during multi-period plnning ( c (), c (), h () ). Compring the lues o the opertor's proit during the single-period nd multi-period plnning showed tht the proit mrgin under usge o the multi-period plnning is 12% higher thn under the single-period plnning. CONCLUSIONS 778

6 Dmytro Ageye et l., Sch. J. Eng. Tech., 2014; 2(5B): In this rticle, we presented n optimition model or the LTE RAN network plnning problem. The mjor physicl nd rdio-electricl prmeters re identiied nd represented by the decision ribles o suitble mixed integer liner progrmming. In order to get the optimition model o multi-period plnning o LTE RAN it is needed to introduce set o prmeters depended on the period number (such s cost o instlltion nd usge o equipment, requirements; prmeters tht relect the current conigurtion o the network) into the single-period plnning model; dd singleinstlltion conditions nd condition tht equipment is not ctited prior to its instlltion. Experiment's results nlysis showed tht ormultion o the problem, which is shown boe nd presented s problem o MILP, llows to obtin the correct solutions rom the prcticl point o iew. We lso need to modiy the objectie unction o the optimition model to ccount or chnges oer time in the cost prmeters. Compring the opertor's proit during the single-period nd multi-period plnning showed tht usge o multiperiod plnning increse the proit mrgin to 12% thn the single-period plnning. REFERENCES 1. Rumney M; LTE nd the Eolution to 4G Wireless: Design nd Mesurement Chllenges.Wiley, 2013: Abdul Bsit, Syed; Dimensioning o LTE Network: Description o Models nd Tool, Coerge nd Cpcity Estimtion o 3GPP Long Term Eolution rdio interce. Msters Thesis submitted in Helsinki Uniersity o Technology, 2009: MichPióro, Deepnkr Medhi; Routing, Flow, nd Cpcity Design in Communiction nd Computer Networks. Elseier, 2004: Yn Zhng;WiMAX Network Plnning nd Optimition (1st ed.).auerbch Publictions, Boston, M US 2009: Andrews JG, Ghosh Muhmed R; Fundmentls o WiMAX: Understnding Brodbnd Wireless Networking (Prentice Hll Communictions Engineering nd Emerging Technologies Series).Prentice Hll PTR, Upper Sddle Rier, NJ, US 2007: Intel in Communictions Adptie Modultion (QPS, QAM), Intel White Ppers on WiMAX,