TESTING	&	VALIDATING	
STADIUM	WI-FI
JARED	GRIFFITH
ROBERT	KRUMM	
Assisted	by:	Michal	Zakrzewski
PROJECTS	COMPLETED:
• 4	x	STADIUMS	- 2014	FOOTBALL	WORLD	CUP	BRAZIL
• 8	x	STADIUMS	IN	INDIA	– T20	WORLD	CUP	/	IPL
• 1	x	RUGBY	STADIUM	SOUTH	AFRICA
• OTHERS…
THE	PROJECT
• CUSTOMER	REQUIREMENTS… 2.4Gz	vs	5GHz	- 70%	on	2.4??
• VALIDATING	A	DESIGN…
• What	are	you	validating	for?
• When?
• Retry	rate	/	Utilization	/	Duty	Cycle?
• Killing	broadcast	traffic	– ARPs	etc.
• LABELS	LABELS	LABELS
• PHYSICAL	VERIFICATION	/	OPTIMIZATION
• AP	LOCATION	/	TILT
• RSSI	MEASUREMENTS	- -45dBm	or	better
• INFRASTRUCTURE	VALIDATION
• CABLE	TESTING	EN	MASSE!	– using	scripts
• BUILDING	A	CHANNEL	PLAN
• Automated
• APs	On	or	Off?
WHY	VALIDATE?
Client	OS	Type	- INDIA
ANDROID
Unknown
Windows	Mobile	(8)
iOS
Windows	7	/	Vista
Others
20%
26%
53%
CLIENT	OS	TYPE	- BRAZIL
Others
Android
Apple
2.4GHz	/	5GHz	SPLIT?
UPLINK	/	DOWNLINK	SPLIT?
%	CONCURRENCY?
VARIANCE?!
WHEN	ARE	YOU	VALIDATING?
WHAT	ARE	YOU	VALIDATING?	
(IN	A	STADIUM)
• PHYSICAL	VALIDATION
AP	PLACEMENT	/	TILTS
• PHYSICAL	OBSTRUCTIONS	/	PERMISSIONS	PROBLEMS
• BEAMWIDTH:		-3dB?	 -6dB?		 -10dB? -20dB?
• FURTHEST	CLIENT	MUST	STILL	RECEIVE	MINIMUM	RSSI!
• WHAT	IS	MINIMUM	RSSI?
• MEASURED	BY	WHAT?
• AT	LEAST	20dB	HIGHER	THAN	NOISEFLOOR!
AP	LABELLING	/	NAMING!
•BIGCLEAR	LABELS	
• READABLE	FROM	± 10m	/	45	ft ?
• DOES	MAC	/	SERIAL	/	NAME	MATCH	THE	ASSET	REGISTER?
• ± 5	DAYS		FIXING	DISCREPANCIES	HERE…
MINIMUM	RSSI…
• NOISE	FLOOR	DURING	GAME:		±75	dBm
• 20dB	SNR	FOR	HIGHER	MODULATION	SCHEMES
• ± 55	dBm TO	WEAKEST	CLIENT		ß KEEP	YOUR	OLD	ANDROID	/	iPHONE
• ± 45dBm	TO	MACBOOK	AIR
• DON’T	FORGET	BODY	LOSS	CAUSED	BY	CROWD!	(±3dB	per	human	in	LOS)
TRANSMIT	POWER…
HIGHER	SNR
HIGHER	MODULATION	(TO	A	POINT)
SAME	CCA	/	CONTENTION!	(<	ENERGY	DETECT)
SAME	SINR
LOWER	SNR
LOWER	MODULATION
SAME	CCA	/	CONTENTION!
SAME	SINR
CARROT	OF	POWER…
• INFRASTRUCTURE	VALIDATION
CABLES,	MEDIA	CONVERTERS	AND	FIBER:
• Hunt	down	disconnected	/	“flapping”	APs.
• Cable	Validation/Certification	not	always	genuine…
• Ruckus	APs	run	ZAP	/	SPEEDFLEX	endpoint…
• Scripted	ZAP	testing	from	MAC	/	LINUX	Laptop	w	Gigabit	Ethernet.
• Plug	into	Access	Switch	– Test	to	all	– Check	ANY	result	<	500Mbps
RESULTS:
udp ver4172.30.9.9:172.30.9.9 172.30.9.74:172.30.9.74 Thu	Feb	18	14:38:02	2016 762.7
udp ver4172.30.9.9:172.30.9.9 172.30.9.220:172.30.9.220 Thu	Feb	18	15:22:18	2016 799.7
udp ver4172.30.9.9:172.30.9.9 172.30.9.87:172.30.9.87 Thu	Feb	18	14:42:24	2016 800.6
udp ver4172.30.9.9:172.30.9.9 172.30.9.11:172.30.9.11 Thu	Feb	18	14:19:40	2016 801.2
udp ver4172.30.9.9:172.30.9.9 172.30.9.138:172.30.9.138 Thu	Feb	18	15:00:49	2016 801.3
udp ver4172.30.9.9:172.30.9.9 172.30.9.18:172.30.9.18 Thu	Feb	18	14:21:46	2016 803.3
udp ver4172.30.9.9:172.30.9.9 172.30.9.84:172.30.9.84 Thu	Feb	18	14:41:23	2016 804.8
udp ver4172.30.9.9:172.30.9.9 172.30.9.231:172.30.9.231 Thu	Feb	18	15:25:29	2016 807.2
udp ver4172.30.9.9:172.30.9.9 172.30.9.123:172.30.9.123 Thu	Feb	18	14:55:46	2016 808.8
udp ver4172.30.9.9:172.30.9.9 172.30.9.61:172.30.9.61 Thu	Feb	18	14:33:40	2016 810.1
50%
NETWORK	VALIDATION
• VLANS
• POLICY	BASED	ROUTES?
• DHCP,	DNS,	LOADBALANCERS,	FIREWALLS,	SWITCHES	ETC
• CAM	/	ARP	/	DHCP	/	ROUTING	/	NAT	CAPACITY?
• AIR	INTERFACE	VALIDATION
CHALLENGE…
• HOW	TO	TEST	AIR	INTERFACE	IN	AN	EMPTY	STADIUM?
• KEY	TO	BUILDING	/	OPTIMIZING	CHANNEL	PLAN	&	TX	POWER
• REQUIRES	MULTIPLE	CLIENTS	/	MULTIPLE	TEST	RUNS
• PEOPLE	ARE	EXPENSIVE…
• CLIENT	DEVICES	VARY	WILDLY…
• MULTIPLE	SIMULTANEOUS CLIENTS,	MULTIPLE	APs,	SAME	CHANNEL…
SOLUTION:
• PERFORM	LIMITED	CLIENT	TESTING	USING	A	“STANDARD	CLIENT”
• SCRIPT	PACKAGE	THAT	ALLOWS	PARALLEL	IPERF	TESTING
• MODIFIED	RUCKUS	R300	AP	+	BATTERY
• R300	CONVERTED	BY	DEVELOPMENT	CENTER	INTO	CLIENT	MODE
• 2.4GHz	/	5GHz	802.11n
• 1x1:1	/	2x2:2	
• SPECIAL	SSID
RESULTS: #	18	results:
10.10.10.176							61.917
10.10.10.200							59.174
10.10.10.173							56.791
10.10.10.205							56.278
10.10.10.213							54.785
10.10.10.175							54.728
10.10.10.216							54.444
10.10.10.174							54.339
10.10.10.199							53.907
10.10.10.204							53.680
10.10.10.207							50.108
10.10.10.208							48.485
10.10.10.215							46.283
10.10.10.214							45.490
10.10.10.203							42.234
10.10.10.212							37.585
10.10.10.179							37.275
10.10.10.217							35.334
#
#	Total	Mbps:						902.838
#
roberts-MacBook-Air:netcap robertkrumm$	./run_iperf.sh -s	staiplist_WestStand_L2
20160329151331,10.10.10.1,52004,10.10.10.213,5001,4,0.0-20.0,137101312,54785271
20160329151351,10.10.10.1,52038,10.10.10.215,5001,4,0.0-20.0,115736576,46283223
20160329151411,10.10.10.1,52063,10.10.10.200,5001,4,0.0-20.0,147980288,59174383
20160329151431,10.10.10.1,52081,10.10.10.205,5001,4,0.0-20.0,140771328,56277761
20160329151451,10.10.10.1,52105,10.10.10.179,5001,4,0.0-20.0,93192192,37274709
20160329151511,10.10.10.1,52143,10.10.10.217,5001,4,0.0-20.0,88342528,35334043
20160329151532,10.10.10.1,52165,10.10.10.212,5001,4,0.0-20.0,93978624,37585018
20160329151552,10.10.10.1,52196,10.10.10.174,5001,4,0.0-20.0,135921664,54338545
20160329151612,10.10.10.1,52228,10.10.10.207,5001,4,0.0-20.0,125304832,50107541
20160329151632,10.10.10.1,52254,10.10.10.203,5001,4,0.0-20.0,105644032,42234404
20160329151652,10.10.10.1,52284,10.10.10.173,5001,4,0.0-20.0,142082048,56790944
20160329151712,10.10.10.1,52322,10.10.10.204,5001,4,0.0-20.0,134217728,53680083
20160329151732,10.10.10.1,52355,10.10.10.176,5001,4,0.0-20.0,154927104,61917205
20160329151752,10.10.10.1,52389,10.10.10.199,5001,4,0.0-20.0,134873088,53907066
20160329151812,10.10.10.1,52422,10.10.10.208,5001,4,0.0-20.0,121372672,48485385
20160329151832,10.10.10.1,52453,10.10.10.216,5001,4,0.0-20.0,136183808,54444071
connect failed:	Operation timed out
20160329152007,10.10.10.1,52489,10.10.10.214,5001,4,0.0-20.0,113770496,45490166
20160329152027,10.10.10.1,52510,10.10.10.175,5001,4,0.0-20.0,136839168,54728224
RESULTS: #	19	results:
10.10.10.209							52.216
10.10.10.213							41.356
10.10.10.205							40.798
10.10.10.176							35.688
10.10.10.175							28.896
10.10.10.204							28.657
10.10.10.217							23.790
10.10.10.174							21.423
10.10.10.212							20.774
10.10.10.173							18.767
10.10.10.215							14.518
10.10.10.207							13.382
10.10.10.208								9.745
10.10.10.214								9.513
10.10.10.216								9.104
10.10.10.203								9.024
10.10.10.200								8.994
10.10.10.199								6.996
10.10.10.179								4.864
#
#	Total	Mbps:						398.504
#
roberts-MacBook-Air:netcap robertkrumm$	./run_iperf.sh -p	staiplist_WestStand_L2
./iperf -c	%s	-y	C	-t 20
20160329152107,10.10.10.1,52578,10.10.10.173,5001,5,0.0-20.0,46923776,18766517
20160329152107,10.10.10.1,52568,10.10.10.213,5001,5,0.0-20.0,103415808,41355645
20160329152107,10.10.10.1,52573,10.10.10.215,5001,5,0.0-20.0,36306944,14517684
20160329152107,10.10.10.1,52579,10.10.10.204,5001,5,0.0-20.0,71696384,28657135
20160329152107,10.10.10.1,52577,10.10.10.217,5001,5,0.0-20.0,59506688,23790185
20160329152107,10.10.10.1,52569,10.10.10.205,5001,5,0.0-20.0,102105088,40798140
20160329152107,10.10.10.1,52586,10.10.10.175,5001,5,0.0-20.0,72351744,28896258
20160329152107,10.10.10.1,52571,10.10.10.212,5001,5,0.0-20.0,52035584,20774411
20160329152107,10.10.10.1,52580,10.10.10.176,5001,5,0.0-20.0,89391104,35688497
20160329152107,10.10.10.1,52583,10.10.10.216,5001,5,0.0-20.0,22806528,9103905
20160329152107,10.10.10.1,52575,10.10.10.179,5001,5,0.0-20.1,12189696,4863580
20160329152107,10.10.10.1,52570,10.10.10.200,5001,5,0.0-20.1,22544384,8993580
20160329152107,10.10.10.1,52574,10.10.10.207,5001,5,0.0-20.1,33554432,13381831
20160329152107,10.10.10.1,52584,10.10.10.214,5001,5,0.0-20.1,23855104,9512630
20160329152107,10.10.10.1,52576,10.10.10.174,5001,5,0.0-20.1,53739520,21422712
20160329152107,10.10.10.1,52581,10.10.10.199,5001,5,0.0-20.1,17563648,6996484
20160329152107,10.10.10.1,52582,10.10.10.208,5001,5,0.0-20.0,24379392,9744602
20160329152107,10.10.10.1,52572,10.10.10.203,5001,5,0.0-20.1,22675456,9024454
20160329152142,10.10.10.1,52585,10.10.10.209,5001,5,0.0-20.0,130678784,52215585
SOME	ADVICE…
• GET	THE	BATTERIES…
FINAL	VALIDATION
• IN	EVENT	TESTING	/	OPTIMIZATION
• ROAMING	OPTIMIZATION	- SMARTROAM
• SPECTRUM	/	PROTOCOL	ANALYSIS	DURING	EVENT
• THROUGHPUT	TESTING	ON	PHONES
• ASK	THE	CROWD!
• STATS	&	REPORTS
• Client	Average	Throughput:
0
5
10
15
20
25
30
35
40
45
5pm	- 6pm 6pm	- 7pm 7pm	- 8pm 8pm	- 9pm 9pm	- 10pm 10pm	- 11pm
Average
Outdoor
PressBox
Stand	A
Stand	B
Stand	C
Stand	D
Stand	E
Stand	G
Stand	M
Stand	N
Stand	P
• QUESTIONS?

WLPC_EU Budapest 2016 | Robert Krumm | Testing and Validating Stadium Wi-Fi