Michael Asmussen 3/20/2018 Introduction

 Motivation: RF systems are perceived as difficult to plan, install, and operate. No metrics used to qualify reliability.  Goal: Attempt to make RF deployments a predominately engineered solution. Insights from 7 years of RF backbones

 Notion of QUALITY has always been subjective – no metrics!

 RF Site Surveys (“Ping Test” are KEY but don’t paint the whole picture)  SNEEZER SURVEY FOR CROWDED AREAS!  No 3D models during deployment  Missing visualization of obstructions or patterns  3D modeling IS MORE than a marketing feature Insights from 7 years of RF backbones

 Case Studies have shown that root causes of Deep Mesh problems are often beyond a technology limitation

 Cleveland University (radio in drip pan, not wifi)  Grand Hyatt Washington DC (Saflok config)  Terranea (Pan merge/size) Finding weak RF backbone areas A way to MEASURE RF quality

 FLQ = Full Link Quality  A measure of systems availability based on total engineered rooms (this measures installation progress).

 CLQ = Conditional Link Quality  A measure of communication quality for all rooms currently communicating (this describes environmental aspects) FLQ and CLQ Definition

 RFLQ / RCLQ (BLUE): RECEIVED (Down Link)

 CFLQ / CCLQ (YELLOW): CONTROL (Up Link)

 TFLQ / TCLQ (GREEN): TOTAL (R X C) FLQ and CLQ Definition

 LOW RFLQ = Received Full Link Quality  Installers or Customer Service should be consulted to improve FLQ

 LOW RCLQ = Conditional Link Quality  An Application Engineer or an experienced / well-seasoned DMN installer should be consulted about ways to improve the CLQ Applying the Available Tools

 3D Hotel Renderings BEFORE INSTALL BEGINS

 Down Link Quality for daily assessment of DMN

 needs IC3 running to accumulate data ○ (data is considered perishable)

 Needs IC3 Version 3.10.5 (Feb/2014) or later Applying the Available Tools

 C:\Inncom\Scripts\PROPERTY_NAME\Policies.cfg ; Netstat feature=142,1,0,0,0,0 Conrad NYC Conrad NYC Conrad NYC Conrad NYC Grand Del Mar San Diego Grand Del Mar San Diego Hyatt Regency Capitol Hill Hyatt Regency Capitol Hill Hyatt Regency Capitol Hill Hyatt Union Square, NYC Hyatt Union Square, NYC Hyatt Union Square, NYC JW Marriott, DC JW Marriott, DC Marriott Marquis Atlanta Marriott Marquis Atlanta Marriott Marquis Atlanta Marriott Sand Key Clearwater, FL Marriott Sand Key Clearwater, FL Sofitel Philadelphia Sofitel Philadelphia Sofitel Philadelphia Hyatt Union Square Hyatt Union Square Hyatt Union Square Hyatt Union Square Renaissance Baltimore Renaissance Baltimore Hilton San Diego Bayfront Hilton San Diego Bayfront Hyatt Regency Waikiki Hyatt Regency Waikiki Sheraton Denver

 2 Buildings  Tower  Plaza  1300 rooms  CELS (Ving) Sheraton Denver Sheraton Denver Sheraton Denver Sheraton Denver Sheraton Denver Sheraton Denver Sheraton Denver Sheraton Denver Sheraton Denver

 But why the difference?

 Tower PAN’s ○ 20 rooms average  Plaza PAN’s ○ 40 rooms average!

 Does it matter? Sheraton Denver Sheraton Denver Building Benchmark Metrics

A start on benchmarking Hotel FLQ CLQ

Marriott Sand Key Clearwater, FL 99% 99%

W LA 96% 99%

Four Seasons Atlanta 98% 98%

Grand del Mar San Diego 98% 98%

George Hotel, DC 97% 98%

Hyatt Union Sqr NYC 98% 98%

New School, NY 98% 98%

Sofitel, LA 86% 98%

W Victory Park Dallas, TX 98% 98%

Hyatt Regency Capitol Hill 96% 96%

JW Marriott, DC 87% 96%

Westin La Paloma Tucson, AZ 83% 96%

Marriott Marquis Atlanta 89% 95%

Conrad NYC 87% 94%

Sofitel Philladelphia 72% 92%

W Union Square, NYC 83% 91%

Westin Chicago River North 69% 74% Building Benchmark Metrics (RCLQ) SA - Building Benchmark Metrics (RFLQ) Project Goals for a proven good DMN

 Strive for 95% or better on RCLQ  Strive for a matching RFLQ once the property is fully commissioned  (those 2 metrics should match at the end)  Reads on those metrics are good after at least 4h of IC3 accumulating data. ○ Remember, it’s starts from scratch when IC3 is restarted. RCLQ Drivers

PAN Size

 Smaller is better, bigger is better too! PAN SIZE

 There is a sweet spot for pan size…  Unfortunately this part is not really an exact science… each project is different.  Several environmental factors that are hard to predict PAN SIZE

 Think 3-dimensionaly!

 Some thumb rules:  Properties with CELS ○ Max PAN 40 ~ 50 ○ (flexible numbers!)  Simple EMS properties ○ Up to 200 rooms!  The more 3rd parties you add, the more data the mesh needs to process.

 Our “safe bet” is 35 rooms per pan from a budgeting perspective RCLQ Drivers

RF Obstructions

 Close proximity to WI-FI AP  rule is Edge Router must be at least 5 ft away!  Water, Water, Water  Piping (risers)  METAL  Cable trays  Rebar  Bunker type construction  Hurricane areas RCLQ Drivers

COMMISSIONING

 Each radio in the room helps with the mesh’s strength for both in-room an backhaul traffic  In-room radios mesh around Room ID  For radios to help with the network mesh they need to all have the same PAN ID!  Commonly network commissioning happens after in-room so RF devices do not have a PAN ID set RF Propagation is Omnidirectional RF Obstructions…

… think SHADOWS

Shaded area due to obstruction

Metal / Thick Concrete

Edge Router RF Obstructions RF Obstructions RF Propagation Indoors No need to alternate RF CHANNELS

 Just go with CH26 everywhere! ○ Unless asked to use another channel or when trying to avoid cross-talk with another hotel ○ Don’t forget to survey if in an “INNCOM Saturated Area”. When in doubt, please survey! It saves lots of $$$$ down the road…  Co-existence of 802.15.4 (ZigBee) and 802.11 (WiFi)

Applying the tools… a few ideas

 Please be diligent about Sneezer Site Survey for neighboring properties  High Quality problem we have!  Augment deep mesh network sign-off process based on the RCLQ and RFLQ metrics  Utilize benchmark numbers as acceptance criteria (are we done? 95% or better!)  Works for RF, RS485 and Ethernet backbone networks!