shanghai breakout: location analytics – key considerations and use cases
TRANSCRIPT
2CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
Today’s Agenda
• Goal: Understand the value of location
analytics for enterprises and public
venues
• And how Aruba ALE together with key
partner solutions can help with various
analytics use cases and drive business
value
3CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
Understanding Analytics
4CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
Location Based Services in Enterprises
• Location Context and
Traffic Pattern Analytics
is becoming
increasingly important
across enterprises and
public venues to
support various
optimization efforts and
mobile engagement
with contextually
relevant information
5CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
Why Location Data Matters
• Improve User/Customer Engagement
– Add context to customer purchase patterns
– Targeted engagement based on location
– Improve Ad effectiveness by > 2X
• Improve Operational Efficiencies
– Staffing Efficiency – Don’t wait for queues
to build – Proactively staff based on traffic
• Workspace Optimization
– Identify “hot zones” or lightly utilized
spaces to save costs
• Location as context for access
control and security
0%
5%
10%
0.10%1.2%
3.5%
7%
10%
Click Through Rate
Source ABI Research
6CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
Big Data Analytics: Market Sizing
7CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
Improve traffic flow
Web analyticsStadium /
Arena
Location Analytics Across Verticals
Increase ad revenues
Optimize traffic flowsAirports /
Malls
A/B Testing
Optimize staffing
Understand buying
patterns
Sentiment analysisRetail
Improve customer
engagement
Identify high value
customers
Real time offersHospitality
Differentiated service
based on location
Workspace optimization
Access policy
management
Enterprises
8CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
Retail Analytics Landscape: Key Trends and Initiatives
SHELF SPACE OPTIMIZATIONCUSTOMER MARKETING
(SEGMENTATION, TARGETING,
PERSONALIZATION)
FRAUD DETECTION &
PREVENTION
INTEGRATED / STATISTICAL
FORECASTING
LOCALIZATION,
CLUSTERING
(DEMOGRAPHIC DATA)
MARKETING MIX MODELING
(A/B TESTING)
PRICING OPTIMZATION PRODUCT
RECOMMENDATION
REAL ESTATE
OPTIMIZATION
SUPPLY CHAIN ANALYTICS;
INVENTORY OPTIMIZATIONTEST & LEARN WORKFORCE ANALYTICS
(STAFF OPTIMIZATION)
MULTI-CHANNEL
ANALYTICS (ONLINE,
OFFLINE)
LOCATION ANALYTICS,
REAL TIME ENGAGEMENTVIDEO ANALYTICS
9CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
Decoding Big Data
10CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
Retail Big Data Topology (Source: IDC, 2012)
11CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
Analytics: Key Takeaways
• Analytics is multi-faceted, complex with
many use cases still evolving
• Several ecosystem players
• Most “real world” implementations require
integration with other data sources
(Sensors, Loyalty databases, POS, etc.) to
create more meaningful data
– May need a SI involvement to put things
together
• Aruba’s ALE provides rich mobility
“context” to analytics and Big Data / mining
systems
• ….but this becomes truly useful only when
combined with multiple data sources to
drive analytics use cases and contextually
relevant user engagement
12CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
An Overview of Aruba Analytics and Location Engine (ALE)
13CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
Mapping LBS Use Cases to Aruba’s Solutions
LBSGuest
Access, Branded Portals
Mobile Engagement
App
Platform
Indoor Mapping Services
Indoor Location
Engine
Contextual Engagement:
Proximity Notifications
Analytics,
Data Mining
ME
RID
IAN
ALE (Network)
Meridian w/BLE
MERIDIAN,
PARTNERS
ME
RID
IAN
CL
EA
RP
AS
SA
LE
+
PA
RT
NE
RS
14CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
Analytics and User / Customer Engagement
Contextual Data:User, Device, Application &
Location
ENGAGEMENTLocation / User Specific
Experiences
DATA
MINING /
ANALYTICS
Sensors
Other
Data
Sources
CRM
Venue Traffic
Patterns, A/B
Testing,
Demographic
Analysis, etc.
ALE
MARKETING, AD
PLATFORMS
15CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
Analytics and Location Engine (ALE): Key Functions
ALE
Unified context for
each user (user name, IP,
MAC, device type, App
visibility, etc.)
1
Seamless, secure
connectivity to
analytics platforms
4
Real time location
engine
2
High performance
Northbound APIs
(publish/ subscribe,
polling)
3
16CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
ALE System Overview
Probing Clients
AP’s Create Virtual
Beacon Report (VBR)
Controllers Create
AMON Messages
ALE imports Visual RF
maps, Decodes AMON,
Computes Location,
Provides Context APIs
ALEAirWaveVisual RF
LOCATION ANALYTICS
Analytics Partner
Location Services
MOBILITY
CONTROLLERS
INSTANT
APs
17CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
ALE Internal Workflow
ALE Processes
Decode the Received data to
appropriate format Location
Engine
Redis In-MemoryDatabase
Calculate Device Location (x,y)
Client RSSI data
Forward decoded User,
Device, App data
North Bound API
Floor Maps from Visual RF (Airwave)
Data from Controller (AMON) or IAP (HTTPS)
Write the received/computed data to DB
Publish the received data using Publish/subscribe API (Google Protobuf/0MQ)
Polling API (REST)
ALE Virtual Machine
18CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
Data Aggregated & Exposed by ALE
• Presence feed
• Indicating a device has been detected in range of the WLAN
• Events when a device is detected crossing a Geofence
• Entering or leaving a zone
• Device information
• Model, OS (as available from DHCP and browser user-agent)
• User information from authentication to the network:
• Type of authentication, username
• Applications used
• As detected by monitoring data-plane traffic from the device
• Destination URLs
• As detected by monitoring data-plane traffic from the device
19CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
Traffic Pattern Analytics Enabled by ALE
Presence (Inside Venues / Conference Rooms)
Capture Rates (Inside versus Walk-Bys)
Dwell Times by Geofence
Repeat versus New Visitors
User Classification (Employees versus Guests)
User Engagement (Macro level)
20CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
ALE Location Calculation Overview
• Location is based on RSSI (from Probes, Data Frames)
– All APs will report RSSI for the probes (Virtual Beacon Report (VBR))
– RSSI from Data Frames (for associated clients) is sent via RTLS feeds
directly from AP’s (or Air Monitors)
• Location calculation based on Path Loss Models
• Path Loss = Received signal – client transmit power
• Path Loss = k + 10 n log(d)
– Where K is the path loss at 1 meter.
– K is different for 2.4 and 5.0 GHz radios.
• If we know the path loss, distance can be estimated
– If we get distance from 3 APs, we can uniquely triangulate
– With 2 APs, there are 2 points of intersection, so there is ambiguity
– ALE returns the AP coordinates (x,y) as proxy to client location when fewer
than 3 AP’s are available for location calculation (“Single AP” location
feature can be enabled via configuration)
• In real life RSSI can fluctuate
– Aruba’s location engine uses outlier detection and dampening algorithms
•
21CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
Location Accuracy & Latency (Summary)
• Factors impacting Accuracy
– AP density, type, mounting type
• Higher the AP (and Air Monitor) density, the better the location accuracy
• Recommended AP / AM density is one every 50 ft (2500 sq ft coverage)
– Client probing behavior, RSSI Variations, Device type, OS type
• Factors impacting Latency
– Client probe frequency (iOS vs Android)
– Network settings: AP/controller timers
• Impact to Use Cases:
– In general, Wi-Fi based locationing from ALE lends itself to use cases where
traffic trends / patterns can be analyzed over a period of time
22CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
Design Considerations for Locationing
• It is imperative to start with a good understanding of business requirements
• What are the key use cases and “true” business requirements?
• Traffic Pattern Analytics inside venues?
• Self directed museum tours?
• Push Notifications by Zone (or with more granularity)?
• Ability to locate specific venue (conference room, restaurant, etc.) within a large venue (statically) or an app that provides turn by turn directions (dynamically)?
• Knowledge of the use case is key to understanding location accuracy, latency requirements – and designing the network to support the use cases
• For “micro-locationing ” or proximity detection and indoor turn by turn direction use cases, a client based solution (BLE) is recommended
23CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
Key Location Analytics Enabled by ALE
Public Venues
(Traffic Patterns,
Engagement)
Enterprise:
Workspace
Optimization
Smart Energy
Management
Integration with
Machine Data
Systems
Location Based
Security PoliciesSDN Enablement
(Context APIs)
24CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
ALE In Action: A Few Case Studies
Analytics Partners
25CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
Hospitality Location Analytics Example (ALE Integration with APAMA)
26CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
Geofence Analytics Example – Hospitality(ALE Integration with APAMA)
27CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
Retail Analytics Example
Market
Benchmarks
Queue
Management
Interior
Counting
Marketing
Effectiveness
Draw
RateShopper
Demographics
Foot
Traffic
= Perimeter Counting
Sales
InterceptLoyalty /
Abandonment
= Mobile Analytics
(Wi-Fi, BLE)
Source: ShopperTrak
28CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
Retail Traffic Analytics Reporting (Sample)
ShopperTrak
Sample Report
(Generated for a
Retail Store in
Spain; integrating
with ALE)
29CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
Retail Traffic Analytics Reporting in Shopping Mall (AisleLabs “Flow” Analytics Sample)
30CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
Traffic Pattern Analysis (AisleLabs Sample Data)
Operations
Information can assist
with planning day-to-
day shopping center
management
operations, such as
staffing
Is a specific marketing campaign effectiveA daily review of peak times will help evaluate and measure the results of promotional campaigns and event programs
• Peak hours remain stable between 10:00 AM - 2:00 PM
Compared to the rest of the Saturdays, guest numbers climbed at 10:00 AM for week #3 and for 6:00 PM for week #4 perhaps due to promotional campaigns.
© 2014 Aislelabs
31CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
Case Study:
Hyper-local marketing campaign (SkyFii)
Turn around a geo-targeted promotion in a short space of time
to drive traffic in-store to redeem an offer.OBJECTIVE u
MEDIA u
RESULTS u
35%conversion
rate
6,067 people received the offer
2,097 were tracked into store
1,642 sales were recorded
Hungry Jack’sCUSTOMER u
Fast FoodCATEGORY u
32CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
Location as Context for Access Policies
Restrict resources by
location for compliance
Restrict guest access to
inside “Geo-fence”
ClearPass
Policy Mgr
Location as
Policy
Definition
Services / SDN
Controller
Device Location
Update / Gepfence
Event
Aruba WLAN
(Access Policy Enforcement based on Location)
XML
API
Dynamic Policy
Update/Enforcement
(CoA)
X
Finger Print
33CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
Machine Data AnalyticsALE – Splunk Integration
Applications
SDK
splunk>
Splunk
Forwarder
Log
Files
Streaming data
DevicesDevices
DevicesALE
Development Kit:
- Interact with the data in Splunk
- Control, manage, script
- SDK support for Perl, Python, Ruby etc.
- Develop custom applications
- 1000s of applications already available
Splunk Engine:
- No RDMS(stored natively)
- Parse/Index/Store the data
- Runs scripts, queries, dashboards
- Cluster & Cloud enabled
- Hunk for Hadoop
- Splunk can be heirarchical (allows distributed
searches)
Data Feed:
- Files & Directories (remote)
- TCP/UDP unstructured data feed
- Forwarders (Universal/Light/Heavy)
- Gather data from network
- Forward (un-indexed) to Splunk Engine
- Compression, SSL, Configurable Buffering
- Feedback from the engine
34CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
Splunk App – Application Visibility Dashboard
35CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
Splunk App – Station Dashboard
36CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
Partner Details
• Expertise: Real time / streaming data analytics
• Focus on Finance industry; new to retail location analytics
• Highly customizable; Integration with other data sources; High cost
• Suitable for large enterprises (e.g. Hyatt Resorts & Hotels)
• Retail foot traffic analytics
• Integration with video camera feeds; other data sources (POS, Loyalty
databases, etc.)
• Customizable reports, alerts; predictive analytics
• Omni-channel KPIs
• Presence Analytics
• Mainly operate in APJ, LATAM, SA
• Standard KPIs: Dwell time, People counts, First Time vs Repeat Visitors, etc.
• Retail and Casual Restaurants (e.g. Westfield Malls)
• Small startup, based in Spain
• Solution focus: Retail Presence Analytics
• Standard Retail Traffic Analytics KPIs: Visitor frequency, Dwell time by zones
• Integration with video feeds
• End to end platform for shopping mall marketing and analytics
• Customizable analytics of shopper behavior
• Social Wi-Fi
• Engagement solutions (with BLE / SDKs)
Key 3rd Party Location Analytics Partners - 1
37CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
Partner Details
• Well know for retail analytics (global list of customers). 20 Year experience
• Started with stereoscopic methods for foot traffic counting; new to Wi-Fi
• integration with other data sources: POS, etc.
• Highly consultative sales / engagement process
• Cloud-based Retail / QSR traffic analytics
• Basic KPIs; some integration with other data sources (POS, etc.)
• Customizable reports including benchmarking, A/B Testing
• Low cost of entry
• Retail traffic analytics; Based in Finland
• Standard KPIs: Engagement; dwell times; identifying loyal customers, etc.
• APIs to external marketing software, Google Analytics, etc.
• Recently acquired by Brickstream
• Started with Wi-Fi only solution (Like Eulid)….now have Beacons for
Engagement, and integration with video feeds for people counting
• Similar store analytics KPIs as others (dwell times, paths, etc.)
• Business intelligence for workspace optimization
• Can integrate multiple data sources (Wi-Fi, secure card readers, other
sensors)
• Predictive analytics
Key 3rd Party Location Analytics Partners - 2
38CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
Summary: Analytics – A Journey
1
2
3
Identify Key Use Cases,
Business Value
Proposition
Tune Network, Identify
Key Partners for POC,
Design Use Cases
POC – 2 to 3 months
Refine Use Cases
4
Build Internal Processes to
consume and act on the data.
Refine Use Cases
39CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
Summary: Key Purpose of ALE
• Context Aggregation and Export
– User, Role, Device, Location, Application
– Meta Data: [URL, Session]
– Real Time Traffic Flows
• ….To Drive key business use cases:
– Traffic Pattern Analytics in Retail and other enterprises
(Presence, Dwell Times by zones, etc.)
– Network / IT Analytics
– Location context for access / security policy
management
• ALE is NOT
– An “indoor Navigation” / “Blue Dot” solution
– A solution for proximity engagement requiring less than
5 m accuracy
A.L.E
40CONFIDENTIAL
© Copyright 2014. Aruba Networks, Inc.
All rights reserved#AirheadsConf
ALE: Key Resources
• Detailed API Document
• Sample Feed Reader Code (0MQ) in C and Java
• Source Code for “ALE Demonstrator App” (Android) on
GitHub
– Shows how to consume both REST and 0MQ APIs
• Help with API programming
• Streaming Data from ALE server for Adapter development
(from Sunnyvale LAB)
• Help with POCs
• …Whatever help you need, we are available!