September 2013 doc.: IEEE 802.11-13/1123r0 Capacity Simulation of High Density Indoor WLAN Systems Date: 2013-09-16 Authors: Name Company Anders Furuskär Ericsson AB Stephen Rayment Ericsson AB Leif Wilhelmsson Ericsson AB Submission Address Phone Färögatan 6 Stockholm, Sweden 168 40 603 March Road Kanata, ON, Canada K2K 2M5 Scheelevägen 23 Lund, Sweden Slide 1 email +46 107 143 679 ander.furuskar @ericsson.com +1 613 254 7070 stephen.rayment @ericsson.com +46 706 216 958 leif.wilhelmsson @ericsson.com Furuskar et al, Ericsson AB September 2013 doc.: IEEE 802.11-13/1123r0 Abstract Preliminary results from simulation of high density indoor WLAN systems. The static simulation is similar to one used for cellular applications Submission Slide 2 Furuskar et al, Ericsson AB September 2013 doc.: IEEE 802.11-13/1123r0 Topics • • • • WLAN multi-cell performance aspects Models and assumptions Results Summary Submission Slide 3 Furuskar et al, Ericsson AB September 2013 doc.: IEEE 802.11-13/1123r0 WLAN Multi-cell Performance Aspects • 802.11n and 802.11ac have similar physical layer characteristics as cellular systems, e.g. LTE – Modulation, channel coding, MIMO capabilities • But… • The CSMA-based MAC limits densification – Nodes that hear each other cannot be active at the same time – This creates a re-use in the time domain • Operation in uncontrolled environments with external interference will further limit performance – Not modeled here Submission Slide 4 Furuskar et al, Ericsson AB September 2013 doc.: IEEE 802.11-13/1123r0 Models and Assumptions • IEEE 802.11n modulation and coding rates • 2.4 or 5GHz • Frequency reuse varied between 1 and 12 vs isolated AP – Choose channels that maximize distance between APs that use the same frequency • • • • • • • • One single 20MHz carrier per AP AP and STA power 100mW (EIRP) CS (Carrier Sense) threshold -85dBm 2x2 MIMO in DL and UL No external interference File transfer traffic 75% in downlink and 25% in uplink MAC PDU size 65KB Submission Slide 5 Furuskar et al, Ericsson AB September 2013 doc.: IEEE 802.11-13/1123r0 Deployment Scenario • An office-like building with 80x50m footprint • Ten floors of height 3m • Interior walls every 10m (10x10m rooms) • Winner II propagation – Free-space propagation with inner wall loss of 12dB, and floor loss of 17dB for first floor and 4dB for following floors – Log-normal shadowing with standard deviation 8dB • Subscriber density 1/16m2 250/2500 per floor/building – ~30 subscribers per AP • 8 APs per floor, randomly deployed per floor, all floors same Submission Slide 6 Furuskar et al, Ericsson AB September 2013 doc.: IEEE 802.11-13/1123r0 Traffic Load per AP vs Monthly Volume per User • Subscriber density 1/16m2 250/2500 per floor/building • 200 busy hours per month • Traffic per subscriber and month v busy hour traffic per floor – Assume 10 GB per month per sub x 250 subs per floor / 200 busy hours per month = 27Mbps per floor or 3.4Mbps per AP Submission Slide 7 Furuskar et al, Ericsson AB September 2013 doc.: IEEE 802.11-13/1123r0 Radio Characteristics and Performance Measurement Method Submission Slide 8 Furuskar et al, Ericsson AB September 2013 doc.: IEEE 802.11-13/1123r0 AP to STA Radio Characteristics • Good coverage – Well above total gain of -100dB which corresponds to -80dBm received power 100 100 90 90 80 80 70 70 60 60 C.D.F. [%] C.D.F. [%] • Good isolation geometry due to walls and floors 50 40 50 40 30 30 20 20 2.4GHz 5.0GHz 10 0 -110 -100 Submission -90 -80 Total Gain (Prx-Ptx) [dB] -70 10 0 -10 -60 Slide 9 2.4GHz 5.0GHz -5 0 5 10 15 20 Downlik Wideband SINR (Geometry) [dB] Furuskar et al, Ericsson AB 25 September 2013 doc.: IEEE 802.11-13/1123r0 Cell Isolation • Propagation as a function of distance and floor • Total loss <105dB (received power >-85dBm) some two to three walls or floors away 2.4GHz 130 130 120 120 110 Floor 9 Floor 8 Floor 7 Floor 6 Floor 5 Floor 4 Floor 3 Floor 2 Floor 1 Floor 0 100 90 80 70 60 50 0 5 Submission 5.0GHz 140 Total Loss [dB] Total Loss [dB] 140 10 15 20 Distance [m] 25 30 110 Floor 9 Floor 8 Floor 7 Floor 6 Floor 5 Floor 4 Floor 3 Floor 2 Floor 1 Floor 0 100 90 80 70 60 35 Slide 10 50 0 5 10 15 20 Distance [m] 25 30 Furuskar et al, Ericsson AB 35 September 2013 doc.: IEEE 802.11-13/1123r0 Cell Isolation Received Power [dBm] • An active AP triggers carrier sensing (>-85dBm) in 7.1 and 5.9 neighbor APs (on average) in 2.4GHz and 5.0GHz respectively -40 2.4GHz x • The graph shows 5.0GHz x -50 x average x -60 received power x x from neighbor APs x -70 in order x x x -80 x x of strength x x x -90 x x x x -100 x 0 2 6 4 8 10 Cell Submission Slide 11 Furuskar et al, Ericsson AB doc.: IEEE 802.11-13/1123r0 Performance Measurement Method • Cell-edge (5th percentile) and mean user throughput vs traffic load • ‘How much can the system be loaded before user experience gets Datarates decrease unacceptable?’ Downlink when load increases due to increased interference and more active users sharing channel 100 Downlink 80 2.4 [100%] m ean 90 80 User Throughput [Mbps] Traffic load 70 C.D.F. [%] 2.4 [100%] 5th perc 70 60 50 40 30 60 50 40 30 20 20 Traffic load 10 10 0 0 20 Submission 40 60 User Throughput [Mbps] 80 100 Slide 12 0 0 1 2 3 4 5 6 Served Traffic per Small Cell [Mbps] 7 Furuskar et al, Ericsson AB 8 September 2013 doc.: IEEE 802.11-13/1123r0 Performance Results 2.4GHz Submission Slide 13 Furuskar et al, Ericsson AB September 2013 doc.: IEEE 802.11-13/1123r0 Isolated APs, 2.4GHz, DL • 5th percentile throughput is at most 40Mbps • Utilization approaching 75% (CSMA efficiency) Downlink 90 Isolated [100%] m ean Isolated [100%] 5th perc User Throughput [Mbps] 80 70 60 50 40 30 20 10 0 Submission 0 10 20 30 40 Served Traffic per Small Cell [Mbps] Slide 14 50 Furuskar et al, Ericsson AB September 2013 doc.: IEEE 802.11-13/1123r0 Reuse 3, 2.4GHz, DL • 5th percentile throughput is at most 12Mbps • Utilization approaching 30% Downlink 90 Isolated [100%] m ean Isolated [100%] 5th perc Reuse 3 [100%] m ean User Throughput [Mbps] 80 Reuse 3 [100%] 5th perc 70 60 50 40 30 20 10 0 Submission 0 10 20 30 40 Served Traffic per Small Cell [Mbps] Slide 15 50 Furuskar et al, Ericsson AB September 2013 doc.: IEEE 802.11-13/1123r0 Reuse 1, 2.4GHz, DL • 5th percentile throughput is at most 7Mbps • Utilization approaching 10% Downlink 90 Isolated [100%] m ean Isolated [100%] 5th perc User Throughput [Mbps] 80 Reuse 3 [100%] m ean Reuse 3 [100%] 5th perc 70 Reuse 1 [100%] m ean Reuse 1 [100%] 5th perc 60 50 40 30 20 10 0 Submission 0 10 20 30 40 Served Traffic per Small Cell [Mbps] Slide 16 50 Furuskar et al, Ericsson AB September 2013 doc.: IEEE 802.11-13/1123r0 Reuse 1, 3, Isolated, 2.4GHz, UL • Same trends as in downlink Uplink 90 Isolated [100%] m ean Isolated [100%] 5th perc User Throughput [Mbps] 80 Reuse 3 [100%] m ean Reuse 3 [100%] 5th perc 70 Reuse 1 [100%] m ean Reuse 1 [100%] 5th perc 60 50 40 30 20 10 0 Submission 0 5 10 15 Served Traffic per Small Cell [Mbps] Slide 17 20 Furuskar et al, Ericsson AB September 2013 doc.: IEEE 802.11-13/1123r0 Capacity • Maximum served traffic for cell-edge (5th percentile) user throughput of at least 5Mbps Downlink Uplink Isolated[100% [100%] Isolated ] Isolated [100% [100%]] Isolated Reuse33[100% [100%] Reuse ] Reuse 33 [100% [100%]] Reuse Reuse 1 [100% ] Reuse 1 [100% ] 0 5 10 15 20 25 30 Served Traffic per Small Cell (at R5=5Mbps) [Mbps] Submission 35 0 Slide 18 2 4 6 8 10 Served Traffic per Small Cell (at R5=5Mbps) [Mbps] Furuskar et al, Ericsson AB 12 September 2013 doc.: IEEE 802.11-13/1123r0 Spectral Efficiency • Capacity divided by total spectrum usage in used direction – Spectral Efficiency = Capacity / (reuse * 20MHz x f) – Downlink = 0.75 Uplink = 0.25 • Increases with tighter reuse, but still rather low Downlink 0 Uplink Isolated [100% ] Isolated [100% ] Reuse Reuse33 [100% ] Reuse Reuse3 3 [100% ] Reuse 1 [100% ] Reuse 1 [100% ] 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 Served Traffic per Small Cell (at R5=5Mbps) [bps/Hz] Submission 0 Slide 19 0.1 0.2 0.3 0.4 0.5 Served Traffic per Small Cell (at R5=5Mbps) [bps/Hz] Furuskar et al, Ericsson AB September 2013 doc.: IEEE 802.11-13/1123r0 Performance Results 5GHz Submission Slide 20 Furuskar et al, Ericsson AB September 2013 doc.: IEEE 802.11-13/1123r0 Reuse 1-12, Isolated, 5GHz, DL • Same trend as in 2.4GHz • Slightly better than 2.4GHz due to better isolation Downlink 80 Isolated [100%] m ean Isolated [100%] 5th perc 70 Reuse 12 [100%] m ean User Throughput [Mbps] Reuse 12 [100%] 5th perc 60 Reuse 6 [100%] m ean Reuse 6 [100%] 5th perc Reuse 3 [100%] m ean 50 Reuse 3 [100%] 5th perc Reuse 1 [100%] m ean Reuse 1 [100%] 5th perc 40 30 20 10 0 Submission 0 10 20 30 40 Served Traffic per Small Cell [Mbps] Slide 21 50 Furuskar et al, Ericsson AB September 2013 doc.: IEEE 802.11-13/1123r0 Reuse 1-12, Isolated, 5GHz, UL • Same trends as downlink Uplink 80 Isolated [100%] m ean Isolated [100%] 5th perc 70 Reuse 12 [100%] m ean User Throughput [Mbps] Reuse 12 [100%] 5th perc 60 Reuse 6 [100%] m ean Reuse 6 [100%] 5th perc Reuse 3 [100%] m ean 50 Reuse 3 [100%] 5th perc Reuse 1 [100%] m ean Reuse 1 [100%] 5th perc 40 30 20 10 0 Submission 0 2 4 6 8 10 12 Served Traffic per Small Cell [Mbps] Slide 22 14 16 Furuskar et al, Ericsson AB September 2013 doc.: IEEE 802.11-13/1123r0 Capacity 5GHz • Maximum served traffic for cell-edge (5th percentile) user throughput of at least 5Mbps Downlink Uplink Isolated [100% [100%]] Isolated Isolated [100%] Isolated [100% ] Reuse12 12 [100% [100%]] Reuse Reuse 12 Reuse 12 [100%] [100% ] Reuse 6 [100% ] Reuse 6 [100% ] Reuse 3 [100% ] Reuse 3 [100% ] Reuse 1 [100% ] Reuse 1 [100% ] 0 5 10 15 20 25 Served Traffic per Small Cell (at R5=5Mbps) [Mbps] Submission 30 0 Slide 23 2 4 6 8 10 Served Traffic per Small Cell (at R5=5Mbps) [Mbps] Furuskar et al, Ericsson AB 12 September 2013 doc.: IEEE 802.11-13/1123r0 Spectral Efficiency • Capacity divided by total spectrum usage in used direction – Spectral Efficiency = Capacity / (reuse * 20MHz x f) – Downlink = 0.75 Uplink = 0.25 • Increases with tighter reuse, but still rather low Downlink 0 Uplink Isolated [100% ] Isolated [100% ] Reuse 12 [100% ] Reuse 12 [100% ] Reuse 6 [100% ] Reuse 6 [100% ] Reuse 3 [100% ] Reuse 3 [100% ] Reuse 1 [100% ] Reuse 1 [100% ] 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 Served Traffic per Small Cell (at R5=5Mbps) [bps/Hz] Submission 0 Slide 24 0.1 0.2 0.3 0.4 0.5 Served Traffic per Small Cell (at R5=5Mbps) [bps/Hz] Furuskar et al, Ericsson AB September 2013 doc.: IEEE 802.11-13/1123r0 Next Steps • Inter-cell carrier sensing based only on AP-to-AP channel, need to look at STA as well • Include hidden nodes • Include RTS/CTS – Currently not explicitly modeled, but no hidden nodes assumed either • Model more realistic modulation rate algorithms • Model effects of more realistic AP channel selection and placement Submission Slide 25 Furuskar et al, Ericsson AB September 2013 doc.: IEEE 802.11-13/1123r0 Summary • High capacity multi cell systems can be built with WLAN technology (!) • Inter-cell carrier sensing limits performance • Possible to mitigate with more channels • Are there more spectrally efficient solutions? • We have a good baseline to assess enhancements Submission Slide 26 Furuskar et al, Ericsson AB September 2013 doc.: IEEE 802.11-13/1123r0 References [1] I. Siomina, A. Furuskär, and G. Fodor. A mathematical framework for statistical QoS and capacity studies in OFDM networks, Proc. of IEEE PIMRC ’09, Sep. 2009. Submission Slide 27 Furuskar et al, Ericsson AB
© Copyright 2026 Paperzz