Dr Michael Capone Principal Analyst - Capgemini
The data generated by IoT-enabled machines, vehicles and devices can provide companies with insight into user behaviour that they can use to create a personal connection with their customers. Companies are, therefore, scrambling to implement IoT systems in order to generate, capture, protect, and analyse this valuable data. But the insights created are only valuable when they trigger consequent decisions and timely actions. There are many potential users of IoT data such as marketing, sales, held service, product
development, customer support, operations, and supply chain not to mention external users like vendors and partners. Each user group needs to be able to access and select different data and apply different logic and analytic approaches to perform specific tasks.
Furthermore, each group can have unique usability requirements. As companies become more IoT mature and start to plan for “data actionability,” the disadvantages of a homogenous IoT stack or departmental systems become obvious. The best option from a data quality, user acceptance, and ROI perspective is a microservices IoT platform.
10. IoT-Cloud
State Engine
2. Machine Data
5. Alert & Task
Technician
ServiceCloud
Manager
Predictive Model
DB
R
BDLaaS
4. Case
1. Input Stream
PLC Data
Machine Data
Apama
Complex Event Processor
3. Descriptive Stream
EBS
WebMethods
Input Orchestrations
2. Complex Stream
Streaming
Manual
Case
Management
Q: How do I identify bad quality during or before the process?
A: PQM
11. IoT-Cloud
State Engine
2. Machine Data
9. Alert & Task
Knowledge
Technician
7. Model + Error Code
8. Article
ServiceCloud
Manager
Assets
6. Model
5. Location + Error Code
Predictive Model
DB
R
BDLaaS
4. Case
1. Input Stream
PLC Data
Machine Data
Apama
Complex Event Processor
3. Descriptive Stream
EBS
WebMethods
Input Orchestrations
2. Complex Stream
Streaming
Manual
Case
Management
Q: How do I prevent unplanned downtime?
A: PMM
12. IoT-Cloud
State Engine
2. Machine Data
9. Alert & Task
Knowledge
Technician
7. Model + Error Code
8. Article
ServiceCloud
Assets
6. Model
5. Location + Error Code
Predictive Model
DB
R
BDLaaS
4. Case
1. Input Stream
PLC Data
Machine Data
Apama
Complex Event Processor
3. Descriptive Stream
EBS
WebMethods
Input Orchestrations
2. Complex Stream
Streaming
Manual
Case
Management
Q: How do I ensure service is performed correctly?
A: PMM w/ VR
13. Customer: Industry: Offering:Departments:
SolutionSituation/Challenges Benefits/Results
Team size:
Project effort:
Project duration:
Q: How do I predict and optimize milk production?
A: Prescriptive Analytics
Lely Agriculture Milking Embedded Software
Development of regression test tool
enabling two types of verification.
Verification of the algorithm output of
the ported code against the original
mathlab results.
Verification of the algorithm output of
a baseline algorithm against the
results of a new algorithm.
New multistage vision algorithms are
developed in mathlab for new
camera’s
Code must be ported to an ARM based
platform running a Linux
distribution.
Fast on-target results of animal health
supporting early indication of performance.
Shorter iterations for evaluation of new
algorithm stages
Objective evaluation of algorithm improvements
achieved over time
33. 1. „Smart“ things are just IoT silos.
Hub
Things can be
connected, but
that does not
mean they are
smart.
Hopper
34. 2. Many disconnected silos.
Hub Bridge
Manufacturers
want to protect
their data and
their customer
relationship.
The more smart
things we
acquire, the more
silos we have to
subscribe to.
Hopper Grinder
35. 3. Frustation.
Hub Bridge EBS
The insights from
one machine
cannot be used to
control another
machine.
Hopper Grinder Mixer Chiller
36. 4. Chaos.
Hub Bridge EBS Gateway
The insights from
one machine
cannot be used to
control another
machine.
IF vacuum cleaner
on, THEN
television louder.
This is not really
smart and
certainly not
sustainable. Hopper Grinder Mixer Chiller Grinder
37. 5. Shadow silos serve other users.
Hub Bridge EBS Gateway
There are many
users for device
data.
To give additional
users access to the
device data, we
typically copy a
subset of the data
to another system
and build a new
application.
Hopper Grinder Mixer Chiller Grinder
TechManagement
38. 6. Valley of Disillusionment
Hub Bridge EBS Gateway
When many
users want
access to the
same data, but
for different
purposes, we
create a
shadow farm.
This is not
sustainable.
Hopper Grinder Mixer Chiller Grinder
Tech
ManagerController Operator HR
39. 7a. Aggregation at gateway
Hub Bridge EBS Gateway
To perform
correlative
analytics and add
external data, a
shadow system
and a parallel
stream are
created,
External data
Hopper Grinder Mixer Chiller Grinder
40. 7b. Replace OEM gateways
Industrial Gateway
or the OEM
gateway is replaced
by one of the
dozen available
multi-protocol
gateways (for
industry, business
and consumer
applications),
External data
Hopper Grinder Mixer Chiller Grinder
41. 7c. Connect multiple databases.
Hub Bridge EBS Gateway
or the
databases
are
integrated.
These
options
enable
intelligence.
This is not
actionable
intelligence. Hopper Grinder Mixer Chiller Grinder
42. 8. Making data actionable.
Hub Bridge EBS Gateway
Intelligence and
actionability are
enabled, but users
have to use multiple
applications. This is
not user friendly.
External data
Hopper Grinder Mixer Chiller Grinder
Tech
Tech Operations Compliance
43. 9. Micro-Services Platform
EBS
Goals
• Any Device
• Any Input Source
• Collaborative Intelligence
• Single Point-of-Truth
• Actionability
• User Friendliness
• Any User
• Any Device
Hopper Grinder Mixer Chiller Grinder
Technologies
1 Enterprise Bus
1 Data Lake
1 Streaming Analytics Engine
1 Enterprise Knowledgebase
1 Process Builder
44. 1o. App
Factory
EBS
Applications for every
user on any device.
Each application can
have ist own
• Logic
• Process
• Dashboard
• UX
defined by the user
group.
Hopper Grinder Mixer Chiller Grinder