Five	Years	of	Research	and	Innovation	
Experience	in	IoT
Lessons	learnt,	DOs	and	DON’Ts	
14th December	2016	
IEEE	WF-IOT	Winter	Academy	on	Internet	of	Things
Raffaele	Giaffreda
Chief	IoT Scientist
Twitter:	@giaffred
outline
•a	layered	perspective	on	IoT challenges
•focus	on	some	key	research	/	business	areas
•turning	research	into	concrete	solutions
•are	we	ready	for	business?
WHO AM I ?
• Chief	IoT	Scientist	- CREATE-NET,	Italy
• 20yrs	experience	in	the	telecom	domain:	BT	and	Telecom	
Italia
• large	projects,	patent	holder,	public	speaking
• >5mEur	funding	acquisition
• IEEE	IoT	newsletter	editor-in-chief	
• MSc, Telecoms	Engineering,	University	College	London,	U.	
of	London
• MSc,	Electronic	Engineering, Optical	Telecommunication	
Systems,	Politecnico	di	Torino
4
About	me
Information	Digital	World
Real	World	of	“information”
100101101100010011
110101101010001010
100101101100010001
101001101010001010
100101101100010000
101101001010001011
100101101100010011
110101101010001010
100101101100010001
101001101010001010
100101101100010000
101101001010001011
100101101100010011
110101101010001010
100101101100010001
101001101010001010
100101101100010000
101101001010001011
100101101100010011
110101101010001010
100101101100010001
101001101010001010
100101101100010000
101101001010001011
100101101100010011
110101101010001010
100101101100010001
101001101010001010
100101101100010000
100101101100010011
110101101010001010
100101101100010001
101001101010001010
100101101100010000
100101101100010011
110101101010001010
100101101100010001
101001101010001010
100101101100010000
100101101100010011
110101101010001010
100101101100010001
101001101010001010
100101101100010000
100101101100010011
110101101010001010
100101101100010001
101001101010001010
100101101100010000
100101101100010011
110101101010001010
100101101100010001
101001101010001010
100101101100010000
100101101100010011
110101101010001010
100101101100010001
101001101010001010
100101101100010000
100101101100010011
110101101010001010
100101101100010001
101001101010001010
100101101100010000
100101101100010011
110101101010001010
100101101100010001
101001101010001010
100101101100010000
100101101100010011
110101101010001010
100101101100010001
101001101010001010
100101101100010000
Real	World Digital	World
100101101100010011
110101101010001010
100101101100010001
101001101010001010
100101101100010000
101101001010001011
100101101100010011
110101101010001010
100101101100010001
101001101010001010
100101101100010000
101101001010001011
THE	IOT	ENABLER
having	something	 to	say…
SENSORS COMMS
EMBED’D
SYSTEMS
PROTOC’S
DATA
STRUCT’S
PLATF’S
transistor	density	/	space efficiency	
Turing’s	Pilot	ACE:	Automatic	
Computing	Engine
TINY
CHEAP
LOW	POWER
Internet	of	***	things
noisy	things
vehicles
smelly	things
radioactive	things
underwater	things
nano things
floating	things
tasty	things
“delle cose belle”
…
RESEARCH	CHALLENGES…
MEMS	(Micro-Electro-Mechanical	Systems)	– see	FBK	J
nanotechnology	
intrabody sensing	for	healthcare	applications
higher	granularity	in	spectrum	of	sensed	entities dot	dot	dot
no	doubt	we	can	sense	/	produce	digital	data	
from	our	real	world	
Real	World Digital	World
100101101100010011
110101101010001010
100101101100010001
101001101010001010
100101101100010000
101101001010001011
100101101100010011
110101101010001010
100101101100010001
101001101010001010
100101101100010000
101101001010001011
giving	voice	to	your	thoughts…
SENSORS COMMS
EMBED’D
SYSTEMS
PROTOC’S
DATA
STRUCT’S
PLATF’S
Cent
MegaHertz
KiloByte
courtesy	of	Mattia Antonini
Constrained	Nodes	(IETF	classification)
Data (RAM) Code (ROM)
Class	0
(Too	constrained)
<<	10	KB <<	100	KB
Class	1
(Quite	constrained)
~	10	KB ~	100	KB
Class	2
(Not	so	constrained)
~ 50	KB ~	250	KB
courtesy	of	Mattia Antonini
IoCT OSes	Features
• Real-Time	Os
• Full	IPv6	Stack
• Multi-hops	support
• Multitasking
• Power	Management
• Application-agnostic
RTOS
Kernel
I/O
Management
Task
Management
Memory
Management
Interrupt	&
Event	
Handling
Timer
Management
Synchronizati
on &
Communicatio
n
courtesy	of	Mattia Antonini
courtesy	of	Mattia Antonini
6LoWPAN
RPL
IPv6
UDPCoAP
CBOR
Flexible
Memory
Management
High
resolution
timersMulti-Threading
Multi-platform
8	– 16	– 32	bits
courtesy	of	Mattia Antonini
RESEARCH	CHALLENGES
• …getting	more	and	more	crammed	into	RTOS…
• energy	efficiency
• size
making	sure	one	can	hear	another…
SENSORS COMMS PROTOC’S
DATA
STRUCT’S
PLATF’S
EMBED’D
SYSTEMS
bandwidth	/	spectral efficiency
The	physics…
• Radio	signal	attenuation	proportional	to	frequency		
• Longer	wavelength,	longer	range	
• Sub-1GHz	band	
• robust	and	reliable	communication	with	low-power	budgets
• bandwidth	limitation
• Modulation	techniques
• (U)NB	vs.	Spread	Spectrum
wireless	technologies	for	M2M
“horses	for	courses…”
LoRa™ Alliance White Paper © Mobile Experts, 2015
BLE	– Bluetooth	Low	Energy
LPWA	– Low	Power	Wide	Area
RPMA	– Random	Phase	Multiple	Access
indoor	 coverage,	low	cost,	long	battery	life	and	large	number	of	devices	(>10K	per	AP)
trendy	comms for	IoT:	LPWANs
• LoRa (non	managed)
• SIGFOX	(managed	services)	
• Vodafone	and	Huawei	(NB-IOT	– 3GPP	LTE	standard)
A	reasonably	well	designed	technology
different	spreading	factors	
(12)	for	different	data	rates
LoRa WIDE	
COVERAGE
not	every	human	
being	runsa as	fast	
as	Usain	Bolt!
LORA	TRENTO	– WORK	IN	PROGRESS
LoRa basic	features
• 868	MHz
• 125	KHz	channel
• 250	bps	– typical
• non-managed
• star	topology,	thousands	of	nodes	/	gateway
• LoRaWAN L2	protocol	for	networking	(security,	duplication	etc.)
SIGFOX
• UNB	– 100	Hz
• 12	bytes	per	message
• 140	messages	per	day	max	(ISM	bands	regulation,	1%	duty	cycle)
• 100	bps
• 2-way	communication
• very	high	power	efficiency
• 1	Eur /	year
NB-IoT
• the	telco	operators’	bet
• 3GPP	LTE
• announcement
• piggybacking	existing	infrastructure
• low-cost	to	deploy,	wide	coverage,	but	
• subscription	based,	quality	league
• Prototypes	exist	but	no	commercial	hardware	/	deployments	yet
5G	anyone?
• issue	of	latency
• tactile	internet	scenarios
• bandwidth…
• but…not	only	radio	technologies
• 5G	is	not	just	about	speed and	more	flexible	networks!
• 5G	is	about	having	a	better	mobile	network	that	can	lead	to	
improved/futuristic	application	smart	scenarios
• 5G	will	in	fact	leverage	on:
• Virtualised/programmablehigh	speed	dynamic	access	&	transport	
networks
• Decreased	latency	thanks	to	Mobile	Edge/Fog	computing	(Tactile	
Internet,	Enhanced	Virtual	Reality,	Telerobotics,…)
• Secure	and	interoperable	IoT	infrastructures	for	a	huge	variety	of		Smart	
Scenarios	(Industry	4.0,	Smart	Cities,	Connected	Cars,…)
things	to	remember	about	5G…
34
RESEARCH	CHALLENGES?
• cheaper
• energy	efficient
• longer	range
• higher	bandwidth
• low	latency
• …
• some	little	extras	(positioning)
don’t	speak	all	at	the	same	time…
SENSORS PROTOC’S
DATA
STRUCT’S
PLATF’S
EMBED’D
SYSTEMS
COMMS
6LowPAN,	CoAP,	MQTT	etc.	protocol	
adaptations	to	optimise the	use	of	wireless,	
low	power,	limited	proc	power…
THIS	IS	ABOUT	GETTING	THE	MOST	OUT	OF	THE	COMM	MEDIUM
TCP	to	optimise use	of	“Best	Effort	Internet”…
…an	example	from	Z-Wave,	
home	automation	protocol…
Research	Challenges
5G	(?)	for	Tactile	Internet	reducing	latency	below	ms
understanding	 the	contents…
SENSORS
DATA
STRUCT’S
PLATF’S
EMBED’D
SYSTEMS
COMMS PROTOC’S
…preparing	gathered	data	to	be	
exploited	by	the	application…
From	standards	to	bespoke	data	structures
• develop	applications	once,	deploy	many	times
• no	additional	coding	for	adding	new	sensors…provided	they	all	sing	
from	the	same	standard	sheet
• about	semantic	interoperability
the	UNCAP	example
channel
stream
Timestamp
POSIT’N
Timestamp
MEASURM’S
Timestamp
ALARMS
Payload
"properties": {
"blood_glucose": {
"allOf": [
{
"$ref": "#/definitions/unit_value”
},
{
"properties": {
"unit": {
"enum": [
"mg/dL",
"mmol/L”
]
}
}
}
]
}
i.e.	location	stream
x,y,z channels
Research	challenges
RESEARCH	CHALLENGES…
easy	learning	books…
SENSORS PLATF’S
EMBED’D
SYSTEMS
COMMS PROTOC’S
DATA
STRUCT’S
what	is	a	platform?	
• a	comprehensive	(software)	offer	of	services	that	puts	together	a	mix	
of	what	presented	so	far
• main	purpose	for	IoT platforms	is	to	provide	more	or	less	automated	
features	that	help	easily	create	applications	that	exploit	data	for	a	
purpose
In	the	case	of	IoT a	platform	will	consist	of…
source:	IoT Analytics
doi:10.1016/j.comcom.2016.03.015
…this	is	also	where	it	starts	to	get	more	crowded!
Open	source
PaaS	vs.	SaaS
Security
Discovery
Remote	management
Interoperability
Supported	 standards
what	makes	a	platform	a	good	one?
whose	chestnuts	do	we	pull	out	of	the	fire?
Facebook Platform open API made it possible for third-party developers to
create applications.
src:	http://www.digitaltrends.com/features/the-history-of-social-networking/
AppleStore Android	GooglePlay
Software	advances
(Hardware	enablers)
touch	screens
tablets	/	smartphones
mobile	computing
Rather than offering a comprehensive social networking experience like the now-defunct
Myspace and the struggling Google+, they instead specialize in a specific kind of
interaction service that involves the sharing of public images (Instagram), the private
sharing of images sharing (Snapchat), augmented reality (Foursquare), and location-
based matchmaking (Tinder). People essentially use the various services in conjunction
with other platforms to build a comprehensive, digital identity.
what	is	the	target?
ease	of	use	for	its	intended	audience!!!
ability	to	tinker	and	personalise it!!!!
contextual	background	awareness…
a	more	comprehensive	picture
• IoT and	Cloud	(infrastructure)
• Industrial	IoT (specific	segment)
• Cognitive	IoT (data)
• Secure	IoT
SENSORS
PLATF’S
EMBED’D
SYSTEMS
COMMS
PROTOC’S
DATA
STRUCT’S
IoT &	
Cloud
Secure
IoT
Industrial
IoT
Cognitive	
IoT
IOT	PLATFORM	AS	A	SERVICE
IoT &	
Cloud
Secure
IoT
Industrial
IoT
Cognitive	
IoT
Improve	IoT through	Cloud
• constrained	devices
• limited	processing	power
• limited	battery	power
• limited	networking
• limited	storage
• limited	support	for	scalable	applications
• advances	in	cloud	computing	(edge	/	fog	computing,	containers,	micro-
services)
constrained	to	unconstrained	offload,	separating	concerns…
two	CREATE-NET	examples
Integration	API
Raptorbox
Problem	addressed
• Challenges	for	integration	of	IoT devices	
into	existing	product/service	portfolio:
• Complexity	of	integration	of	heterogeneous	
IoT devices	into	an	existing	infrastructure:
• Interaction	with	IoT devices	(device	identification,	
protocol	handling)
• Security:	secure	communication,	device	and	data	
access	control
• Scalability:
• From	few	devices	in	trial	phase	to	massive	
deployment	of	IoT connected	devices
• How	to	perform	rapid	prototyping	to	address	
fast	business	and	tech	validation	cycles	and	
fast	delivery
Service	Bus
Enterprise	Systems
Device	integration	and	
management	made	eas
in	a	secure,	scalable,	
configurable	way		
courtesy	of	Fabio	Antonelli
Our	solution
u Device	Virtualization:
u Common	Device	Modeling	(“Web	
of	Things”	paradigm)
u IoT	Message	Brokering:
u Scalability	by	design
u Multiprotocol	support	(http/https,	
MQTT,	JMS,	AMQP)
u Data	chaching for	real-time	event	
processing	and	querying
u Configure	your	Business	Logic	for	Rapid	IoT	Application	Prototyping	(Data
and	events	workflow	Editor)
u Flexible	Access	Control	&	Authorization	(ACLs)	for	devices	and	users
u Secure	Communication	and	Interaction	with	devices
u Easy	Integration	via	APIs exposing	all	available	capabilties
courtesy	of	Fabio	Antonelli
Integration	API
Raptorbox Service	Bus
Enterprise	Systems
the	Raptorbox IoT Data	Broker
COMMUNICATIONS
SENSING
GOOD	DATA
VALUE	GENERATION
ROUTING
FILTERING
the	more	I	understand	the	data,	
the	better	value	I	can	provide…
AGGREGATING
INTERPRETING
VALUE
PROCESSING
low	 high
low	
high
store	significant	data…
Payload
"properties": {
"blood_glucose": {
"allOf": [
{
"$ref": "#/definitions/unit_value”
},
{
"properties": {
"unit": {
"enum": [
"mg/dL",
"mmol/L”
]
}
}
}
]
}
“literate”
(relevant	plugins	/	libraries)
Raptorbox IoT
Data	Broker
higher	processing	but…
save	storage	space
facilitate	interpretation
save	network	use
“all	blood	glucose	levels	above	a	threshold”
Raptorbox target
• system	integrators	mainly
• focus	on	core	service	provisioning	competences	
while	exploiting	interoperable	platform	for	enriching	
those	with	interoperable	IoT data	harvesting
• examples:	SMEs	digitalisation support,	smart	cities,	
e-health	
Integration	API
Raptorbox
Service	Bus
Enterprise	Systems
WAZIUP	Platform
The	EU-AFRICA	WAZIUP	platform	(Actor	view)
App.	Development
App.	Deploy
Sensor	registration
App.	Execution
Developer
Sensor	owner
App	user
Third	party	API	
integration	
Data	provider
courtesy	of	Corentin Dupont
App source
code
data
processing &
analytics
IoT PF IoT sensors
Architecture
courtesy	of	Corentin Dupont
WAZIUP	target
• African	community	of	developers
• focus	on	core	competences	while	exploiting	ready-to-use	open-source	
tools	and	components	to	cater	for	the	needs	of	African	businesses
• examples:	fish	farming,	precision	agriculture,	cattle	rustling	etc.	
App source
code
data
processing &
analytics
IoT PF IoT sensors
research	challenges
Dynamic	instantiation	of	IoT
functions	(microservices)	 on	
edge	cloud	infrastructure
Cloud4IoT,	D.	Pizzolli et	al.
Setting	the	scenes	on	Industrial	IoT
IoT &	
Cloud
Secure
IoT
Industrial
IoT
Cognitive	
IoT
Image	Credit:	The	Industrial	Internet	Consortium	– April	2015	Infographic
TREND:
we can	sense	and	transmit	more	
and	more	efficiently
why	do	we	want	to	do	that	in	
an	industrial	context?
recent	news…
• software	industry
• appliance	/	electronics
• “SAP	and	Bosch	team	up	on	Internet	of	Things”
• …	The	technology,	for	example,	allows	a	production	system	to	select	
the	torque	for	each	screwdriver's	task,	increasing	efficiency ...	
• wow...what	does	it	take	to	tighten	a	screw?	
• how	much	torque	to	apply	when?	what	about	replacing	the	
screw	driver?	what	about	ensuring	it	is	the	right	one	for	the	
type	of	screws?
• sensing	system	and	an	actuator...plus	contextual	knowledge	
about	type	of	screw,	screw	pitch	and	size,	material	(pre-
sales)...data	collection,	interpretation	(after-sales)
enhance	a	particular	task
integration,	know-how
enhance	a	particular	task
Raccoltadati
Descriptive
what	happened?
Diagnostic
why	did	it	happen?
Predictive
what	will	happen?
Preventive
what	should	I	do?
Decision
Actuation
Decision	support
Decision	automation
human	input	requiredanalytics
TREND:	servitization
(sell	&	forget	vs.	sell	and	assist)
enhance	a	particular	task
all	well,	but…
• need	reliable	technology
• sensing	and	communications
• security,	dependability,	servitization
• (pre-sales	/	after-sales)		
• need	performance
• data-processing	and	edge	cloud
• need	competences	(choice,	integration,	deployment)
• infrastructure	
• choice	of	technologies
• interface	between	standards
• middleware
• flexible	architectures
• interface	between	standards
• services	and	applications
• what	knowledge	do	we	want to	extract	from	data?
• interface	between	standards
• need	technologists	+	domain	experts,	working	side	by	side
enhance	a	particular	task
need	reliable	technology
• (sensing)	– what	to	sense,	size,	durability,	etc.
• securely	getting	data	out	of	sensors	to	the	applications
• what	options	for	your	production	plant,	assembly	line,	deployment	
environment…
• 5G	is	a	key	enabler	
• reliable	communications	/	protocols
• energy	efficiency
• short	round-trip	delays
• NB-IoT vs.	Sigfox vs.	LoRa
need	performance
appsense sense appprocess
vs.
edge	cloud	/	fog	computing
need	competences		
• know-how!
• infrastructure	
• choice	of	technologies
• interface	between	standards
• middleware
• flexible	architectures
• interface	between	standards
• services	and	applications
• domain	experts	+	technologists
• solution	design	
integration,	know-how
Advantages	of	4th industrial	revolution
• digitalisation of	production	process
• digitalisation of	product
• monitoring	during	and	after	production
• azienda manifatturiera acquisirà componente da	software	house
• “just	in	time”	production	– with	management	of	stock,	stores,	
production	value	chain
• products	personalisation
• reduced	production	and	final	product	costs	– competition
• new	business	models	tied	to	servitisation
BREAK	TIME?
MAKING	SENSE	OF	HARVESTED	IOT	DATA
IoT &	
Cloud
Secure
IoT
Industrial
IoT
Cognitive	
IoT
http://waitbutwhy.com/2015/01/artificial-intelligence-revolution-1.html
TED	Talk	on	AI
• https://www.ted.com/talks/sam_harris_can_we_build_ai_without_lo
sing_control_over_it
• https://www.ted.com/talks/jeremy_howard_the_wonderful_and_ter
rifying_implications_of_computers_that_can_learn
• dave evans https://youtu.be/UQi9B7yB5dM (beginning	of	it	size	of	
sensing	+	computing)	phone	processing	power	in	a	blood	red-cell…
• deep	learning:	computer	can	see	(10min	30)	also	they	see	what	we	cannot	
(heartrate	from	image...),	computer	can	read
unlocking	a	huge	potential
data
data
data data
data
data
data
data
data
data
data
data
data
data
H/W
presence
patterns	exist	...SENSING
data	goldmine	
and	lots	of	
siloed
applications
presence
derive patterns	of	...
interpret	
data
Extracting	knowledge	from	data	– domain	
expert	modeling…
• many	bespoke	machine-learning	applications	exist
• however,	still	substantial	overhead	needed
• smart-agriculture	example
• domain	expert	models	need	to	assist	machine	learning	experts	to	help	them	
design	algorithms	that,	based	on	collected	data,	can	actuate	according	to	
model	expectations
• sometimes	models	need	to	be	created	through	observation	(lengthy	process)
• in	both	cases,	a	lot	of	validation	data	is	needed	to	train	and	tweak	algorithms
• no	wide	applicability,	no	general	purpose	machine	learning…
• experience	from	iCore EU	collaborative	project
how	cognitive	technologies	and	
IoT can	be	leveraged	upon	to	
optimise network	resource	
usage	in	a	smart-city	security	
monitoring	application
Alcatel	Lucent	Bell	Labs	/	Thales
courtesy	of	Marc	Roelands
Illustration by Justin Metz
applicable	in	all	
scenarios	exposing	
highly	structured	data
the	emergence	of	unsupervised	learning…
(+	advances	in	edge	cloud	computing)
SECURITY	AND	PRIVACY	IN	IOT
IoT &	
Cloud
Secure
IoT
Industrial
IoT
Cognitive	
IoT
IoT and	Security
https://www.pentestpartners.com/blog/new-wi-fi-kettle-same-old-security-issues-meh/
MIRAI	DDOS	ATTACK	– October	2016
many	levels	of	security
• data	encryption	at	transmission	level
• data	encryption	at	storage	level
• policy-based	access	control
• anonymise data
• etc.
• IoT and	blockchains…(enable	secure	and	logged	exchange	of	IoT
messages)
Added	Value	for	IoT	
• Trust	and	Reputation	of	IoT	devices:
• Non-Repudiable Device	Identity
• Security	enforcement	at	the	edge
• Secure	Traceability	of	Transactions	and	of	Information:
• in	financial	transactions,	supply	chains,	and	other	processes	involving	involving	IoT	devices
• transparency,	auditability	without	the	need	to	leverage	on	3rd party	trusted	entities
• Make	consumer	data	more	private
• More	Resiliency:
• No	single	point	of	failure
• IoT	devices	can	autonomously	interact	with	humans	and	other	IoT	devices:
• including	capabilities	to	perform	automatic	payments/value	exchange	tracking	(digital	
currencies)
courtesy	of	Fabio	Antonelli
SENSORS
PLATF’S
EMBED’D
SYSTEMS
COMMS
PROTOC’S
DATA
STRUCT’S
IoT &	
Cloud
Secure
IoT
Industrial
IoT
Cognitive	
IoT
IOT	SYSTEM	INTEGRATION	CHALLENGES
transforming	research	ideas	into	concrete	
solutions
• market	need:	remote	monitoring	of	elderly	and	
cognitively	impaired	people	(through	wearables,	indoor	
location	etc.)
EU	H2020	Innovation	Action	
in	a	nutshell
Standardised Interfacing	to	
wearables /	monitoring
Common	services	
across	pilots
Applications	to	carers
Applications	to	
patients
Monitoring
Overall	architecture
! ! ! ! ! !!
! ! ! ! ! !UNCAP!BOX!
!
!
!
!
!
User’s!clinical!data!
!
!
!
!
!
!
!
!
!
!
!
UNCAP!BOX!core!
(incl.!seman9c!proxy)!
UNCAP!end>
user!end!point!
UNCAP!
caregivers!end!
point!
Smart!flooring!
–!based!
localisa9on!
Biometric!
sensors!
Camera!based!
localisa9on!
ZigBee!
localisa9on!
Other…!
Local!DB!
(configura9on)!
Wearable!
sensors!
Other!…!
Interoperable!channels!(HL7!/!CDA!/!
SensorThings)!over!secure!connec9onx!
UNCAP!services!
in!the!cloud!
Outer&world&
Value!added!services!
from!3°!par9es!
Public!health!IT!services!
(based!on!standards)!
Proprietary!!
data!format!
Loca9on!
Connectors!
Sensor!
Connectors!
Home&or&&
protected&&
environment&
Loca9on!informa9on!!!(users!&objects)!
!
!
!
!
!
!
!
!
!
!
!
Interoperable!
channels!!
Standardised Interfacing	to	
wearables /	monitoring
Common	services	
across	pilots
Applications
Monitoring
The	IoT Interoperability	Problem
• Technical	Interoperability
• ability	to	communicate	between	objects	
or	between	objects	and	apps
• Need	for	a	“Multi-protocol	IoT hub”
• Syntactical	Interoperability
• about	data	formats,	i.e.	xml,	html,	json
• Semantic	Interoperability
• to	do	with	resource	annotation,	query	
and	discovery	(i.e.	semantic-based)
• ontologieshelp	with	common	vocabulary	
and	knowledge	sharing
The	SMEs	real	world	problem…
• “semantic	what?”
• understand	scalability,	RESTFUL	API	exposed	services
• understand	the	need	to	implement	access	control	and	safe	storage
• use	of	IoT Data	Broker	cloud-based	platform
UNCAP	Software	
Beta	Release
Data	
Broker
Cognitive	
and	
Physical
Clinical	
Assess
Location Fencing
Usr
Mgmt
Auth and	
Login
Storage
PHR
SysLog
CEP
CommCache,	
Sync	&	
Dispatch
Device	
Index
Data	
Broker
Data	
Buffer
Location
Video	
Camera
Glucometer
SensFloor
HeartRate
Pulse
Oximeter
Blood	
Pressure
UNCAP	
WebApp
Scale
BLE	
devices
Medical	
Reminders Billing
iHealth
Secure,
Permanent				
Storage
IoT Data	
Broker	
(cloud)
data	sources
data	sources
data	sources
data	sources
data	sources
data	sources
IoT Data	
Broker	
(gateway)
IoT data	(direct)
IoT data	
(via	gateway)
CEP,	data	
processing
access	
control
MQTT,	STOMP,	CoAP,	
REST,	WebSockets
OAUTH
eHealth	scenario	integration
Open	mHealth standard
Offline	functionality
validated	for
DB	integration	(CHINO)
Node-red	based	rapid	
prototyping	 for	business	
logic
deployment	 on	both,	
large	and	small	scale
Integration	API
Raptorbox
Problem	addressed
• Challenges	for	integration	of	IoT devices	
into	existing	product/service	portfolio:
• Complexity	of	integration	of	heterogeneous	
IoT devices	into	an	existing	infrastructure:
• Interaction	with	IoT devices	(device	identification,	
protocol	handling)
• Security:	secure	communication,	device	and	data	
access	control
• Scalability:
• From	few	devices	in	trial	phase	to	massive	
deployment	of	IoT connected	devices
• How	to	perform	rapid	prototyping	to	address	
fast	business	and	tech	validation	cycles	and	
fast	delivery
Service	Bus
Enterprise	Systems
Device	integration	and	
management	made	easy,	
in	a	secure,	scalable,	
configurable	way
applying	the	results	
of	a	project	to	solve	a	
concrete	problemchannel
stream
Timestamp
POSIT’N
Timestamp
MEASURM’S
Timestamp
ALARMS
i.e.	location	stream
x,y,z channels
• Open	mHealth compliance
• “payload”	to	carry	JSON	
structures,	but…
• Value	is	in	someone	else’s	
hands
• Decided	to	accommodate	
more	detailed	structures
• parsing,	searching,	filtering	by	
thresholds	etc.
the	UNCAP	example
channel
stream
Timestamp
POSIT’N
Timestamp
MEASURM’S
Timestamp
ALARMS
Payload
"properties": {
"blood_glucose": {
"allOf": [
{
"$ref": "#/definitions/unit_value”
},
{
"properties": {
"unit": {
"enum": [
"mg/dL",
"mmol/L”
]
}
}
}
]
}
i.e.	location	stream
x,y,z channels
Integration	API
Raptorbox Service	Bus
Enterprise	Systems
the	Raptorbox IoT Data	Broker
COMMUNICATIONS
SENSING
GOOD	DATA
VALUE	GENERATION
ROUTING
FILTERING
the	more	I	understand	the	data,	
the	better	value	I	can	provide…
AGGREGATING
INTERPRETING
VALUE
PROCESSING
low	 high
low	
high
choices	made
• web-of-things	(W3C)	compliance
• JSON	data	structure	native	for	better	value	creation
Securing	access	to	datastreams
IoT Data	
Brokerdata	sources
data	sources
data	sources IoT data	
upload
usr_ID
sess_KEY
data	TYPE,	VALUE,	etc.
User	Mgmt
sess_KEY
LOGIN	
(usr_ID,	pwd)
sess_KEY
UNCAP	Webapp
usr_ID,
nok_ID,	
Dr1_ID,
Dr2_ID
sess_KEY ßà usr_ID ?
1.
2.
3.
CREATE	(sess_KEY ßà usr_ID)
4.
5.
6.
7.
usr_ID
nok_ID
Dr1_ID
Dr2_ID
PUB/SUB	
CHANNELS
8.
outline
•a	layered	perspective	on	IoT challenges
•focus	on	some	key	research	/	business	areas
•turning	research	into	concrete	solutions
•are	we	ready	for	business?
THE	BUSINESS	OF	IOT
WHY	ISN’T	HAPPENING	YET?
where	is	the	IoT?
• no	broad	set	of	applications	encompassing	“one	IoT”
• with	mobile	phones	and	personal	computers	it	was	easier
• IoT devices	very	diverse,	yet	we	tend	to	blur	boundaries
• losing	ability	to	tackle	separately	different	markets
DISCLAIMER:	no	business	expert	but	have	matured	insights	into	the	business	of	IoT that	might	be	useful	to	share
All	IoT examples	but…
smart	locks
thermostats
lights
health
“Home”
power	OK
costs	LOW
“industrial”
power	LOW
costs	No	constraints
SOME	KEY	QUESTIONS
•what	business	model?
•is	this	worth	x	Eur/month…
•to	me?
•to	my	intended	market	audience?
•to	my	public	administration?
Return	on	investment	(ROI)
• EXAMPLE	1
• I	spend	a	$	to	buy	a	bottle	of	water	
because	I	am	thirsty
• the	(immediate)	need	=	I	am	thirsty
• who	benefits?	=	me	(private)
• willingness	to	pay	for	it	=	I	need	it	badly
• when	do	I	benefit	=	as	soon	as	I	get	my	
bottle
• I	make	an	(private)	investment,	the	
benefit	is	immediate
• VERY	SHORT	CYCLE,	TANGIBLE,	
UNAMBIGUOUS,	CONCRETE
B2C
• EXAMPLE	1.b
• I	spend	$	to	buy	an	iPhone
• the	(immediate)	need	=	I	need	a	cool	
device
• who	benefits?	=	me	(private)
• willingness	to	pay	for	it	=	can	do	cool	
things	with	it
• when	do	I	benefit	=	as	soon	as	I	get	it
• I	make	an	(private)	investment,	the	
benefit	is	immediate
• VERY	SHORT	CYCLE,	TANGIBLE,	
UNAMBIGUOUS,	CONCRETE
location	is	key	– booth	next	to	a	fountain? “coolness”	is	key	– no	“cheap	look”	please…
IDENTIFY	YOUR	POTENTIAL	MARKET	TARGET…
Return	on	investment	(ROI)
• EXAMPLE	2
• I	spend	money	to	make	my	house	energy	efficient
• the	(not	so	immediate)	need	=	I	need	to	save	money	on	
my	energy	bills
• the	(good	for	a	common	cause)	need	=	I	need	to	make	
my	life	more	sustainable
• who	benefits?	=	me	(private),	the	environment
• willingness	to	pay	for	it	=	I	need	it	(not	so	badly),	the	
environment	needs	it	(not	so	badly)
• TIME	DIMENSION
• when	do	I	benefit	=	after	I	paid	the	bills	for	needed	
equipment	with	the	money	I	saved
• I	make	an	investment,	the	benefit	might	be	for	someone	
else	or	not	materialise until	later
• LONG-ISH	CYCLE,	TANGIBLE,	UNAMBIGUOUS,	CONCRETE	BUT…
B2G2CB2C
• EXAMPLE	2.b
• smart-lighting
• the	(not	so	immediate)	need	=	I	need	to	save	money	on	
my	energy	bills
• the	(good	for	a	common	cause)	need	=	I	need	to	make	
my	city	more	sustainable
• who	benefits?	=	the	environment
• willingness	to	pay	for	it	=	the	city	balance	sheet	needs	it	
(in	a	couple	of	years,	not	so	badly),	the	environment	
needs	it	(not	so	badly)
• TIME	DIMENSION
• when	do	I	benefit	=	after	I	paid	the	bills	for	needed	
equipment	with	the	money	I	saved
• I	make	an	investment,	the	benefit	might	be	for	someone	
else	or	materialise when	it	is	too	late
• LONG-ISH	CYCLE,	TANGIBLE,	UNAMBIGUOUS,	CONCRETE	BUT…
Return	on	investment	(ROI)
• EXAMPLE	3
• I	have	a	business	and	I	want	to	digitilise it
• spend	money	to	make	my	production	process	more	modern	and	efficient…
• the	(not	so	immediate)	need	=	I	need	to	gain	competitive	advantage
• the	(good	for	a	common	cause)	need	=	I	need	to	gain	insights	into	my	business	operations
• who	benefits?	=	my	biz	(private)
• willingness	to	pay	for	it	=	I	need	it	(not	so	badly),	long-term	gains
• TIME	DIMENSION
• when	do	I	benefit	=	as	soon	as	I	am	in	a	position	to	transform	gathered	data	into	differential	
advantage	that	drives	more	customers	to	buy	what	I	sell	or	reduces	operating	costs	etc.
• I	make	an	investment,	the	benefit	is	not	immediate	and	depends	on	a	proper	strategy	
• LONG	CYCLE,	UNTANGIBLE
B2B2C
The	value	(and	diversity)	of	data
• the	importance	of	bespoke	modeling	– multi-disciplinarity and	
adjacent	domain	experts	interactions
• cycles	of	learning	(modeling)	before	I	can	be	predictive	and	even	
longer	before	I	can	be	prescriptive…
• sensing	and	influence	on	results...
• IS	IT	WORTH	IT?
Example:	motors	manufacturing	 biz
vibration,	current,	torque
MTBF:	60000	hours	(!)
Raccoltadati
Descriptive
what	happened?
Diagnostic
why	did	it	happen?
Predictive
what	will	happen?
Preventive
what	should	I	do?
Decision
Actuation
Decision	support
Decision	automation
human	input	requiredanalytics
the	ROI	CYCLE
market	segmentation
ROI
COSTS
IMPACT
B2C
B2B
B2B2G
SHORT LONG
LOW
LOW
HIGH
HIGH
Descriptive
what	happened?
Diagnostic
why	did	it	happen?
Predictive
what	will	happen?
Preventive
how	to	avoid	it?
build	hindsight
what	insight	
do	I	need?
foresight	and	
optimise
Time
complexity
potential	
gains
California	US	trip
WHO	we	solve	the	problems	for	and	WHY
• WHO
• application	developers	(rapid	
prototyping)
• system	integrators
• system	admin	of	eHealth
• API	framework	managers
u WHY
u rapid	development	saves	costs	&	time
u agility
u easy	integration
u hide	complexity,	Web-based	APIs
9
key	message	– who	is	your	target?
• Cisco	(Jasper),	IBM	(Bluemix),	GE	(Predix)	…
• IoTango,	Trilogis etc.
• propose	a	reference	framework	for	validation	of	how	to	break-down	a	
complex	problem	space	into	more	“palatable”	“mouth-sized”	chunks
up-front	investments	and	ROIs
IS	IT	WORTH	IT?
OPPORTUNITIES	ARE	TREMENDOUS
WARNING!!!
THIS	IS	DAUNTING	IF	YOU	WANT	TO	EAT	IT	ALL
ROI	CYCLE	LENSES	MIGHT	HELP
NEED	TO	BREAK	IT	DOWN	IN	SMALLER	CHUNKS
BUSINESS	MINDSET
Thank	you!
• Linkedin:	https://it.linkedin.com/in/raffaelegiaffreda
• Twitter:	@giaffred

Tutorial at IEEE WF-IOT Dec. 2016 - Five Years of Research and Innovation Experience in IoT