Going Beyond RF Coverage: Designing for Capacity Andrew von Nagy 5 GHz 2.4 GHz 1997 1999 2003 2009 2011 2013 Revolution Wi-Fi
Have you experienced this? + Hint: It s NOT an RF coverage issue
How Many AP s Do I Need?
Goal for Today: Promote the use of capacity planning when designing EVERY WLAN 1. Definition 2. Measurement 3. Methodology 4. Integration
This is NOT a High-Density Session!
WLAN Lifecycle DESIGN OPTIMIZE & GROW DEPLOY TSHOOT & REMEDIATE MONITOR & REPORT
Typical Design Process Predictive Modeling Deployment Site Survey Focus is on signal strength and coverage only
WLAN Requirements Coverage User Mobility Capacity High Availability Wired Integration AAA Services IP Addressing Security Proper Configuration Code Stability Monitoring & Visibility Troubleshooting Tools
What Determines WLAN Capacity? All APs Single Client Multiple Client Coordinated APs
Defining WLAN Capacity Medium Contention RF Interference Net WLAN Capacity (-) (+) Spectral Inventory Spectral Efficiency Goal: Increase spectral inventory and efficiency, decrease medium contention and RF interference
Spectral Inventory (+) 1. Unlicensed Spectrum (by region; growth) 2. Optimal Use of Spectrum: - - - Wider Channels More Radios, Smaller Channels Channel Reuse
Medium Contention (-) 1. Minimize # STA on same channel / AP radio
Medium Contention (-) Optimize Link Efficiency Initial Contention Window size affects performance
Medium Contention (-) 2. Minimize co-channel interference (APs & Clients) Typical WLAN (omni antennas, standard overlap for roaming) Need 4 non-overlapping channels to effectively minimize co-channel interference at -85dBm DFS Non-DFS 20 MHz 22 9 40 MHz 10 4 80 MHz 5 2 160 MHz 1 0 106 138 42 58 58 138 106 138 58 42 58 138 10 United States Regulatory Domain One cell buffer before channel must be re-used
High-Density Channel Planning Always, Always, ALWAYS... use 20 MHz channels! Higher radio density Spectrum re-use harder Clients still create CCI with omni
Spectral Efficiency (+) 1. Packing More bits per MHz (Shannon s Theorem) Modulation, Spatial Streams, MU-MIMO, (Wider Channels - very little gain from addtl subcarriers) QAM-16 QAM-64 QAM-256
Spectral Efficiency (+) 1. Packing More bits per MHz (Shannon s Theorem) Modulation, Spatial Streams, MU-MIMO, (Wider Channels - very little gain from addtl subcarriers)
Spectral Efficiency (+) 2. Client Airtime Efficiency Lowest Data Rate Lower Data Rate Higher Data Rate Highest Data Rate
RF Interference (-) Identify and Eliminate: - Design: Site Survey - Production: Network Monitoring
Measuring Capacity Airtime Utilization
Measuring Capacity Sticky Situation, This Is!!! Application Throughput Frequency Bands Protocol Version Client Mix Airtime Demand Channel Width Signal Quality Spatial Streams
Design Methodology Key: Design for BOTH coverage and capacity! Design Requirements: Device - what type of device(s) are used? Laptops vs. tablets vs. smartphones vs. embedded Identify data rates and actual throughput Application - bandwidth demand, latency req.? Quantity - how many total concurrent devices? Forecasting Capacity: Airtime - shared amongst all devices (in same contention domain) Capacity - dependent on both client and AP capabilities
Method for Forecasting WLAN Capacity 1. Inventory Client Devices & Applications
Method for Forecasting WLAN Capacity 2. Calculate Airtime Consumption per Device
Method for Forecasting WLAN Capacity 3. Forecast AP Capacity
Method for Forecasting WLAN Capacity 3. Forecast AP Capacity Y% airtime utilization (each) Z% airtime utilization (each) X Qty. X Qty. Total airtime utilization based on client mix
Example #1 Scenario: 4 K-12 Classrooms in a Building Wing Devices: 11n tablets (20 MHz, 1SS), 65 Mbps (30 actual) Application: 2 Mbps YouTube Streams Quantity: 30 per classroom (120 total), only 50% concurrent Airtime: 2 Mbps / 30 Mbps = 6.66% Total: 6.66% * 60 devices = 400% = 4 AP radios (2 dual-radio APs)
Example #2 Scenario: 300 Seat University Auditorium Devices: 11n tablets (20 MHz, 1SS), 65 Mbps (30 actual) 11n laptops (20 MHz, 2SS), 144 Mbps (70 actual) Application: 1 Mbps Web / Email / Social Media Quantity: 100 tablets, 200 laptops 100% concurrent
Example #2 Scenario: 300 Seat University Auditorium Airtime: tablets 1 Mbps / 30 Mbps = 3.33% laptops 1 Mbps / 70 Mbps = 1.43% Total: tablets 3.33% * 100 devices = 333% + laptops 1.43% * 200 devices = 286% = 619% = 7 AP radios (4 dual-radio APs)
WLAN Design Process
Grab the Worksheets http://www.revolutionwifi.net/p/downloads.html
Supplementary
Spectrum Inventory Varies by Region (Return)
Spectrum Inventory FCC proposal for additional unlicensed spectrum (Return)