Models for Wireless Infrastructure economics & Mobile Broadband deployment. Outline

Save this PDF as:

Size: px
Start display at page:

Download "Models for Wireless Infrastructure economics & Mobile Broadband deployment. Outline"


1 Models for Wireless Infrastructure economics & Mobile Broadband deployment Jens Zander 1 Outline Some fundamental problems in infrastructure provisioning Wireless Network design fundamentals Wireless Broadband dimensioning & deployment models 1

2 Some fundamental questions 3 The last mile problem: Most investments in Access Networks 1 km Backbone network shared by many Access network individual 1 km 1m Example: cities 5. user each Access network: 1 m/user Trunk network: 1 m/user (=1 km/1. users) 4

3 Wireless operator costs Administration 15% Network Maintenance 14% Annualized Equipment 6% Billing 1% Annualized Capital Costs 16% Marketing 8% Interconnect 9% 5 Wireless Network Dimensioning - a recap 6 3

4 Wireless Networks - problems ange Coverage E N rx P B tx Gant user ( eff ) 7 Peak rates & PHY-technology is no longer THE issue.. Edholms law 8 4

5 ange limitations Typical ranges (NLOS): 1 kbit/s (GSM) 5+ km 5 kbit/s (EDGE) 5-1 km Mbit/s (UMTS/) -3 km 1 Mbit/s (HSPA) 5 m 1 Mbit/s (LTE/WLAN) 5-15 m N BS A A tot cell 1 B G user ant P tx / 9 Wireless Networks - problems cont. Interference due to spectrum reuse Capacity limitation 1 5

6 The infrastructure cost Cinfra c1 cbs NAP cbs NAP Spectrum limitation Btot available bandwidth Spectral /reuse efficiency K B N B A B C c c c infra tot user user user tot user BS BS BS Wsys Wsys Wsys Coverage limitation N BS / 1 N cell P B / user 11 Mobile Data Tsunami Cisco forecast: 15 6x Extrapolation: - 1x 1 6

7 Lowering the system cost B A B C c cw c cw tot tot user sys 3 sys 3 sys Wsys Wsys Improving the efficiency of the modulation and M system, i.e. increasing educing the coverage area A tot. The required data rate is only provided in parts of the area Buying more spectrum? educing the cost per base station c C C C C C C AP site backhaul equipment deployment maint 13 Partial coverage 14 7

8 HET NETS from blanket coverage to selective capacity (Klas Johansson, Cost Effective Deployment Strategies for Heterogeneous Wireless Networks, Doctoral Thesis, KTH 7) 15 Technology shift Consumer grade eq Low power/short range Low system complexity (P&P, SON) Massive deployment eliability through redundancy Industry grade eq High power 4-7 availabilty High system complexity 16 8

9 The cost of spectrum B tot sys sys BS BS BS Btot Wsys NBSWsys WSYS cbs C c N c C More base stations B BN W NW N W tot BS sys sys BS More spectrum C C C c N c N c W sys sys BS BS BS sp B W min C min cbs,cbs NBS csp SYS c c c N Engineering value of spectrum * BS sp BS BS WSYS B NBS Is mobile spectrum still cheap? Source: B G Mölleryd and J Markendahl Valuation of spectrum for mobile broadband services - The case of Sweden and India ITS egional Conference, New Dehli, Feb 1 9

10 Mobile broadband deployment 19 Coverage & Bit rate 1

11 Traffic, $ What can be done about the revenue gap? Traffic Lower cost Increase value/revenue Telcos revenues Voice dominant Data dominant Time In a world dominated by data, traffic growth is not anymore correlated with revenue growth! Design Example: ural deployment 6 cont ( r ) cw log E [ ] W 1 '( r ) f cp tx N r ( r ) dr W ( r ) 5 4 Data rate (cont) EDGE UMTS HS Cell radius: 5 m Peak ate (Mbps) Avg. ate (Mbps) 3 Continous N/A,84 1 HS UMTS.66 EDGE.38.38,1,,3,4,5,6,7,8,9 1 Avg distance 11

12 Design Example: Urban deployment 16 Cell radius: 15 m Data rate (cont) EDGE UMTS HS Peak ate (Mbps) Avg. ate (Mbps) Continous N/A,5 HS 7. 1,9 UMTS 1,6 EDGE ,1,,3,4,5,6,7,8,9 1 Avg distance Design Example: Very dense deployment 35 Cell radius: <5 m 3 Data rate (cont) Peak ate (Mbps) Avg. ate (Mbps) EDGE UMTS HS Continous N/A 5,1 HS 7. 3,8 UMTS EDGE ,1,,3,4,5,6,7,8,9 1 Avg distance 1

13 Single cell capacity & approximation 35 Single cell capacity E[ ] W '( r) f ( r) dr Data rate (cont) HS 1 5,1,,3,4,5,6,7,8,9 1 5 Deployment strategies Wide area blanket coverage Low Capacity 6 13

14 Deployment strategies Limited Hot spot coverage High capacity Capacity 7 Capacity enhancement More spectrum (channels) E[ ] W '( r) f ( r) dr tot N ch Sectorization Improved spatial reuse tot N s 8 14

15 Temporal design peak capacity Networks designed for peak/busy hour 9 Dimensioning For voice and T data you need to estimate the maximum number of ongoing calls or sessions Depends on the arrival rate and the duration of calls Is based on the traffic during the busiest hour For data NT data traffic the approach with average data rate per user can be used X GB per user and month -> Y kbps per user During 4 hrs all day(s) During - 8 hours per day 15

16 Numerical example 1 Gbyte/month = 3 Mbyte/day (= 1.3 Mbyte/h average ) = 4-5 Mbyte/h peak hour ( all daily traffic in 6-8h) = 48Kbyte/36 s = 1.5 Kbyte/s = 1 Kbps Population density: 1 pop/sqkm Cell size: 1.5 m = 6,8 sqm => 68 pop/cell Capacity demand: 1 * 68 =8,5 Mbps /cell => 8,5/3 = Mbps/sector 31 Energy constraints Global scale: Energy consumption of IT-technology not neglectable (% of CO -emission) 3G technology example Base station F output (at antenna): 6 W Power input: 3-6 kw (Efficiency 1-%) eason Spectrum efficient not power efficient ELECTICITY BILL 3. BS = 1 GWh/day = 1 MSEK/day 3 MSEK/month / 1 M Users 3 SEK/month SEK/KWh) 6 SEK/month SEK/kWh) 16

17 What Power to use? 3,5 3,5 1,5 1 Data ate vs Power r=1 r=.8 ( r ) cw cw log 1 cw log 1 log N c ' P N cp r r 1 tot N tot BS N cp tx r,5,1,,3,4,5,6,7,8,9 1 r=.6 r=.4 r=. 1 N BS r 33 What cell size to use? equired total power 1 Mbps Mbps 5 Mbps 1 Mbps c' P cw log 1 N r cw P 1 tot c1 r tot -5,1,,3,4,5,6,7,8,

18 Some conclusions Peak & average data rates differ a lot Cell capacity = Average data for user in cell Increase capacity by more channels & Sectors Dimensioning for peak-hour traffic Total energy consumption decrease with cell size 35 18