School of Electrical and Information Engineering University of the Witwatersrand, Johannesburg Impact of Frequency and Radio Channel Variability on a Converged Wireless Access Network and Telecommunications Service Provision in Sub-Saharan Africa Peter J. Chitamu Centre for Telecommunications Access and Services p.chitamu@ee.wits.ac.za www.ee.wits.ac.za/~comms/ • Introduction Outline – Background on CeTAS Research thrusts – Convergence viewed from Access Network parameters • Presentation of Typical Wireless channel parameters from measurements results – Pathloss model parameters – Multipath model parameters • Examples of applications of the results – Reflections on Coverage (Range) – Reflections on throughput limitation (Rate) • Conclusions 1 CeTAS Research Thrusts How to Create Innovative ICT Applications Using Telco networks? How to Drive the Cost of telecom Access down? How to Create Points of Integration For Connections And OSS? CeTAS How to provide Application-related Terminal-to-terminal Quality of Service Across networks? CeTAS Research Thrusts Services and Applications In an Open Service Access Environment Quality of Service and Networks Performance The Next Generation Network Access Networks And Planning 2 Convergence and Networks LAN LAN VoIP Gateway PSTN Converged Network Model New Billing Server OSS Server Application Connection Control Old Gateways IP Network with QoS SDP Signalling System No 7 SG Web Server SCP TE TG IP PSTN LE MG LE NAS Access DSLAM BB-WLL Campus LAN ADSL Copper Loop WLL 3 A view of Integrated Telecomms. Network Call Collection Area (Access) Local Transport National Transport Global Comms. Voice VoIP Data PSTN Voice VoIP Data Options: 1. cdma2000 fixed - mobile 2. WLL 3. WLAN 4. Broadband Wirele ss Gateway Gateway Local Intranet Internet PSDN Reach (Loss model) & Rate (Bps) Reach / Rate Traffic Source & Sink Models From another node, CN bps, N calls/sec To another Node 1 1 2 2 Node A n Node B Full duplex lines CT=(NCL) bps To another Node m Local loops, CL bps, calls/sec 4 Reach problems in Sub-Saharan Africa (Traffic density in proportion to Population density) Typical Rural and Urban demand 1200 1000 800 600 400 200 0 1 Urban Europe 2 3 4 Cell Radius [km] Urban Africa 5 6 Rural Africa Network Convergence and Services • Access Network to support a wide range of services – Rate Problems (what is average kBps?) – Voice, data and multimedia (10 kbps – 500 kbps) • Access Network to support users in different operating environments – Reach problems (How far subscriber can be?) – Rural Vs Urban – Low density Vs High density • For wireless Access Technologies – Convergence tied to radio channel characteristics and hence Radio Access Network technology 5 Service Type vs Technology Capabilities As Dec 2002 M El-Sayed and J Jaffe,”A View of Telecomms. Network Evolution”, IEEE Comms. Mag., Dec 2002, pp. 74-81 Propagation measurements in Tanzania • Purpose – Obtain channel parameters for Reach estimation – Obtain channel parameters for Rate estimation – Use the models to specify a suitable RAN • Propagation measurement data used – Evaluation of performance of selected network technologies – Recommend features of a suitable Radio Access – In the design of a Rural Radio Systems (DaRT) 6 Propagation Measurements • Experimental Setup – Transmitter – Receiver • Results – Pathloss – Multipath • Model results and its implications Example of Propagation Measurement Equipment: Transmitter 1 3 0 .2 M Hz x x DC P RBS 3 0 0 M Hz 4 3 0 .2 M H z TRAN S CEIVER 5 M Hz 7 Example of Propagation Measurement Equipment: Receiver A T TN . TRANS C E IVE R X LT O P 3 00 M Hz X 1 3 0 .2 M H z DC P RBS 5 M Hz - fC Technical Specs Operating frequency 300 MHz Antennae Half wave dipoles, 11m vertically polarised Transmit power 10 watts (40dBm) PRBS length 511 (29 - 1) Chip rate 5 Mbits / sec Time delay resolution 0.2 usec Time delay range 102.2 usec Time compression ratio 5000 Receiver noise figure Overall around 2dB 8 Propagation Measurement sample sites USA River Propagation Measurement sample sites Kilombero 9 Propagation Measurement sample sites Mwanza Propagation Measurement sample sites Serengeti 10 Propagation Measurement sample sites Moshi, Kilimanjaro Receiver & Transmit test bed 11 Hilly areas in Lushoto Different receiver sites 12 Soni, Lushoto (Families separated by nature) Lushoto, Tanga 13 Different receiver sites Mombo, Tanga 14 Njombe, Iringa Transmit sites and some attractions 15 Multipath selected Results 2.0 1.0 0.0 -1.0 -2.0 -2.5 10220.0 10180.0 10300.0 10260.0 Time samples, PRBS correlator output LU1162.LIN 1.2 Multipath selected Results 1.1 1.0 0.9 -6 dB 0.8 0.7 V o l t s 0.6 0.5 0.4 0.3 0.2 0.1 0.0 -0.1 -2.0 1.0 4.0 7.0 10.0 13.0 16.0 MICROSECONDS 19.0 22.0 25.0 28.0 16 AR1082.LIN Multipath selected Results 1.2 1.1 1.0 0.9 0.8 0.7 V o l t s -6 dB 0.6 0.5 0.4 0.3 0.2 0.1 0.0 -0.1 -3.0 2.0 7.0 12.0 17.0 22.0 27.0 MICROSECONDS 32.0 37.0 42.0 47.0 52.0 Multipath selected Results ar1082.imp 1.2 USA River Arusha, Tanzania 1.0 0.8 0.6 0.4 0.2 0.0 -0.2 - 0 . 00 0 0 1 0 0. 0 0 0 0 1 0 0.000030 0 . 00 0 0 5 0 0. 0 0 0 0 7 0 17 AR1092.IMP Multipath selected Results 1.2 1.0 Pre-Cursors and Post-cursors 0.8 Nearby echos and a group from far 0.6 0.4 0.2 0.0 -0.2 - 0 . 0 00 0 1 0 0 .0 0 0 0 10 Time 0 .0 0 0 0 30 in 0 .0 0 0 0 50 0 .0 0 0 0 70 seconds Multipath selected Results TA1 153.IM P 1.2 A real strange profile 1.0 0.8 0.6 Pre-Cursors Post-Cursors 0.4 0.2 0.0 -0.2 -0.000015 -0.000005 Time 0.000005 in 0.000015 seconds 18 Scatter Plots Pathloss models PL(d) [dB] = PL(d0) + 10 n log (d/d0) where n is path loss exponent & constant PL(d0) from linear regression of scatter plots 19 Path loss exponents from propagation measurements ! Hilly and heavily forested terrain ( Tanga, Lushoto, USA River, Arusha, Kilimanjaro ) P = 72 log d + 88 L 10 dB ! Undulating landscape with light vegetation ( Iringa, Morogoro, Njombe ) P = 53 log d + 93 L 10 dB ! Flat terrain with light vegetation ( Mwanza area ) P = 42 log d + 67 L 10 dB Performance Evaluation of selected RAN • Reach example using derived models – DECT Vs GSM – Frequency Variability • Fading (NLOS) – Flat Vs Frequency Selected – Rate and Reach Implications • Techno-Economic Assessment – TDMA Vs CDMA – Sensitivity Analysis 20 Example 1: DECT and GSM Band Modulation Rate Uplink power (Ave) Equalization DECT GSM (900) 1880 - 1900 Uplink 890 – 915 270.8333 kbps (3.6923 s) 1W / 125 mW 30 dBm / 21 dBm Rural Area, Hilly Terrain, Typical Urban, Test Pattern 1152 kbps (0.868 s) 10 mW (10 dBm) None: Assumed Microcell, office environment Rural Africa ? DECT Power budget -50 -60 -70 -80 -90 -100 -110 -120 -130 -140 -150 -160 -170 -180 2 Area 1 ( 10 dBm) Area 1 ( 24 dBm) Area 2 ( 10 dBm) Area 2 ( 24 dBm) Area 3 ( 10 dBm) Area 3 ( 24 dBm) LogNormal Fading Margin Receiver Threshold 4 6 Distance d [ km ] 8 10 21 GSM Power budget -50 -60 -70 -80 -90 -100 -110 -120 -130 -140 -150 -160 -170 -180 2 Area 1 (13 dBm) Area 1 (30 dBm) Area 2 (13 dBm) Area 2 (30 dBm) Area 3 (13 dBm) Area 3 (30 dBm) LogNor mal Fading Mar gin Receiver Threshold 4 6 8 10 12 14 Distance d [ km ] 16 18 20 Example 2: Impact of terrain and operating frequency Free Space (n=2) Pathloss 80 60 40 20 0 0 2 4 6 8 10 12 14 Distance d [ km ] Free Space (300 MHz) 16 18 20 Free Space ( 900 MHz) Free Space (1800 MHz) 22 Pathloss from Measurements (300 MHz) Flat, Undul & Hilly at 300 MHz 200 180 160 140 120 100 80 60 40 20 0 0 2 4 6 300, 900 & 1800 MHz 8 10 12 14 Distance d [ km ] 16 Free Space ( 300 MHz) Free Space ( 900 MHz) Free Space ( 1800 MHz) Prac Flat 300 MHz Prac Undulating 300 MHz Prac Hilly 300 MHz 18 20 Pathloss from Measurements Impact of terrain 200 180 160 140 120 100 80 60 40 20 0 We found that TDMA and CDMA systems are affected differently by channel variability 0 2 4 6 8 10 12 14 Distance d [ km ] 16 Prac Flat 300 MHz Prac Undulating 300 MHz Prac Hilly 300 MHz Prac Flat 900 MHz Prac Undulating 900 MHz Prac Hilly 900 MHz 18 20 23 Implications on Large coverage areas • High Density Traffic Areas e.g. Cities – Close to the design parameters • Low Density Traffic Areas e.g. Rural Areas – There could be large variations compared to design values – Coverage Limited possibility of outtage PL (d ) = PL ( d 0 ) + 10n log LogNormal Shadowing d + χσ d0 P(out ) = P( PR ≤ Rth ) ( = P ( PT − PL (d ) − L M ) ≤ Rth ) Example 3: Multipath Implications • General Observations – Usually more extensive that typical European cities – Rural areas can have very large delay spread due to NLOS large coverage areas (measured up to 50 microseconds delay spreads) • Implications – Throughput can be severely curtailed due to Multipath induced Inter-symbol Interference – Need a robust Radio Access Network (inbuilt counter-measures against multipath delay spread) – 3G is designed to take advantage of multipath, so one possible option 24 GSM Rural Channel Impulse Response Source: Raymond Steele, Mobile Radio Comms, Pentech Press GSM Urban Channel Impulse Response Source: Raymond Steele, Mobile Radio Comms, Pentech Press 25 GSM Hilly Channel Impulse Response Source: Raymond Steele, Mobile Radio Comms, Pentech Press Delay resolution Imaginary 2 1 Real 0 3 Time (1 sec intervals) Delay resolution ( s) DECT = 0.868 GSM = 3.69 CDMA = 0.814 UMTS = 0.26 Coherent Vector sum 3 3 2 1 0 Real 2 I Delay Resolution II III Delay ( s) 26 Overall Coverage Uncertainty BS MS PT D IS T A N C E L o g n o rm a l S lo w fa d in g PDF S lo w fa d in g M a rg in 1 -2 % R a y le i g h F a s t fa d i n g PD F PR F a st fa d in g M a rg in 1 -2 % Southampton, England 27 Singapore London, England 28 London, England Some village in Lushoto 29 Some villages in Ifakara, Kilombero, Tanzania Urban areas have many options • Urban Reach – – – – Cable Fiber Broadband WLL 3G (UMTS & cdma2000) • Urban Rate – Copper (ADSL) – All of above • Non-Urban Reach – – – – Wireless (GSM) 3G (e.g. cdma2000) Satellite (But expensive) ? • Non-Urban Rate ( > 144kbps) – 3G (e.g. cdma2000) Most service areas in Africa can be classified as Non-Urban 30 Summary • Access Network Capability is a key area in Converged Networks and services – Should provide for both Reach and Rate (variable) • Sub-Saharan Africa has unique conditions – Requirements for large coverage areas/low traffic – Propagation results suggest potential degradation and coverage limitations for systems with low delay resolution – Requires careful selection of Access Network Technology to meet both Reach and Rate requirements of Multi-service platforms concurrently References • • • P. J. Chitamu and D. E. Vannucci, Networking Africa: A Case for VoIP, The Transactions of the SA Institute of Electrical and Engineers, Vol. 94 No: 2, June 2004, pp. 83-90 P. J. Chitamu, I. G. Kennedy, and R. Van Olst, “Technology options to support telecommunications in countries with low teledensities”, 2nd International Symposium on Communications and Info Tech (ISCIT2002), 23-25th October 2002, Pattaya, Thailand.On CD-ROM ISBN 974-537-149-1, pg. 289-292. P. J. Chitamu and P C Chen, “Economic Viability Study of Cellular Network Technologies for Low Density Traffic Areas”, Proceedings of 3rd International Conference on Information (ICICS2001), Communications, and Signal Processing, 15-18 October 2001, Singapore, On CD-ROM ISBN 981-04-5149-0 31