US010733311B2

(12) Chen
United States Patent (10 ) Patent No.: US 10,733,311 B2
et al. (45 ) Date of Patent: Aug. 4 , 2020
(54 ) COGNITIVE INTERNET OF THINGS ( IOT) (56 ) References Cited
GATEWAYS FOR DATA SECURITY AND
PRIVACY PROTECTION IN REAL - TIME U.S. PATENT DOCUMENTS
CONTEXT- BASED DATA APPLICATIONS 5,797,134 A 8/1998 McMillan et al.
8,484,113 B2 7/2013 Collopy et al.
( 71) Applicant: International Business Machines (Continued )
Corporation , Armonk , NY (US )
FOREIGN PATENT DOCUMENTS
( 72 ) Inventors: Sue Ann Chen , Carlton (AU ); Fatemeh
Jalali, Hawthorn East (AU ) ; Mahsa JP 2002259708 A 9/2002
Salehi, Southgate (AU ); Arun
Vishwanath , Blackburn (AU ) OTHER PUBLICATIONS

(73 ) Assignee : International Business Machines Statista , “ Largest insurance companies worldwide as of Mar. 2016 ,
Corporation , Armonk , NY (US) by market capitalization (in billion U.S. Dollars)", Statista, https: //
www.statista.com/statistics/431547/leading-insurance-companies
( * ) Notice : Subject to any disclaimer, the term of this worldwide-by -market-cap /, Accessed on Mar. 27 , 2017 , 4 pages .
patent is extended or adjusted under 35 (Continued )
U.S.C. 154 (b ) by 260 days. Primary Examiner Jeffrey C Pwu
(21) Appl. No.: 15 /472,369 Assistant Examiner William A Corum , Jr.
(74 ) Attorney, Agent, or Firm — Scully , Scott, Murphy &
(22) Filed : Mar. 29 , 2017 Presser, P.C .; Joseph Petrokaitis
(57) ABSTRACT
(65 ) Prior Publication Data
A cognitive IoT gateway may receive real- time data asso
US 2018/0285580 A1 Oct. 4 , 2018 ciated with a user from sensor devices that are detecting user
data in real-time. The real- time data is analyzed to determine
(51) Int. Ci. whether the real-time data indicates a deviation from a
G06F 21/00 (2013.01) predicted user behavior data predicted based on historical
GO6F 21/62 ( 2013.01) data . The real-time data is transformed into collated data that
(Continued ) removes protected and sensitive information from the real
(52) U.S. CI. time data , and the collated data is transmitted to a server
CPC G06F 21/6218 ( 2013.01); G06F 21/316 computer, wherein the protected and sensitive information
received from the sensor devices remain local to the cogni
(2013.01 ); G06F 21/6254 (2013.01) ; G06Q tive IoT gateway. Updated information computed by the
40/08 ( 2013.01 ); H04W 12/02 (2013.01) server computer based on the collated data is received and
( 58 ) Field of Classification Search a real-time alert is sent to a user device notifying the user of
CPC GO6F 21/6218 ; G06F 21/316 ; G06Q 40/08 ; the updated information .
HO4L 63/02
See application file for complete search history . 15 Claims, 5 Drawing Sheets
-202
SERVER PREDICTS USER BEHAVIOUR USING HISTORICAL DATA
DEVICE RECORDS INFORMATION ON CURRENTUSER BEHAVIOUR 204
206
CURRENT USERBEHAVIOUR DEVIATING FROM PREDICTED
NOH ATYPICAL BEHAVIOUR ? YES
LOCAL
GATEWAY ANALYZES AND 208
REVIEWS DATA FROM DEVICE
GATEWAY SENDS REVIEWED AND COLLATED - 210
DATA TO SERVER
PREMIUM -212
DETERMINATION SERVER INITIATES CALCULATION OF
UNIT INSURANCE PREMIUM BASED ON
224 CURRENT CIRCUMSTANCES

SERVER MAINTAINS PREVIOUSLY PREMIUM = FN ( E.G .,WEATHER , - 214

DETERMINED INSURANCE PREMIUM SAFETY ,ACCIDENT RATE ,
TIME OF DAY,LOCATION )
SERVER SENDS UPDATED 216
INSURANCE PREMIUM TO
GATEWAY
LOCAL
GATEWAY SENDS REAL - TIME
ALERTAND PREMIUM UPDATES -218
TO USER'S DEVICE
CHANGES IN USER'S ACTIONS -220
RECORDED BY DEVICE AND
SENT TO GATEWAY

PROCESS IS REPEATED AFTER A DEPINED

DURATION OR F CONTEXT OF USER CHANGES
-222
 US 10,733,311 B2
Page 2

(51) Int. Ci. 2015/0324923 A1 * 11/2015 Christensen G06Q 40/08

G06F 21/31 ( 2013.01 ) 705/4
G06Q 40/08 (2012.01) 2016/0071219 A1 * 3/2016 Joshi G06Q 30/0269
705/4
H04W 12/02 ( 2009.01 ) 2016/0171521 A1* 6/2016 Ramirez G06Q 30/0224
701/409
( 56 ) References Cited 2016/0277528 A1 * 9/2016 Guilaume HO4L 67/306
2017/0365007 A1 * 12/2017 Huls G06Q 40/08
U.S. PATENT DOCUMENTS 2018/0025430 A1 * 1/2018 Perl B6OR 16/0231
705/4
9,162,032 B2 10/2015 Lynch , Jr. 2018/0050698 A1 * 2/2018 Polisson B60W 30/182
2002/0128882 A1 * 9/2002 Nakagawa G06Q 30/02 2018/0075538 A1 * 3/2018 Konrardy G06Q 40/08
705/4
2004/0153362 A1 8/2004 Bauer et al . OTHER PUBLICATIONS
2007/0027726 A1 2/2007 Warren et al.
2013/0297418 A1 11/2013 Collopy et al. Pramanick , S., “ Dynamic Insurance Pricing — Telematics Analytics
2014/0067434 Al 3/2014 Bourne et al. & Behavioural monitoring” , LinkedIn , https://www.linkedin.com/
2014/0257874 A1 9/2014 Hayward et al. pulse/dynamic -insurance -pricing-telematics-analytics-pramanick
2014/0343972 A1 * 11/2014 Fernandes GO8G 5/0039
705/4 fca -pmp, Published on Jun. 1 , 2015 , Accessed on Mar. 27 , 2017, 3
2015/0032481 A1 * 1/2015 Norling -Christensen pages .
G06Q 40/08
705/4 * cited by examiner
U.S. Patent Aug. 4 , 2020 Sheet 1 of 5 US 10,733,311 B2

104
SERVER RDSETLAOVRENAST TOINSURANCE PREIMC UNMGDAETNRMIDES IPNRSEUMAICE PRICES
IRSEVINWED NFORMATION SERVER71USPEDANTED INSURACE PRIECM NUG
TO
SREAL-TIMEDATAENDS LOCAL
G1COATN0EIWTAVYE REAL-TIMEDATASENDS AREVNIEDWS RAENL VYAZNETS DATATOBESENT

10
5

REAL-TIMESAELNDRST
SREAL-TIMEENDS ALERT
PTOARDEJMUISTM SRENADLS TIMEDATA SRENADLS TOTIMEALERT ADJUST PREMIU PTOARDEJMUISTM FIG
1
.

C
OCUPANY DETCIONDEVICES WM
0

ANDIOTDSEVNISCOERS WEARBLS GEORAPHY TRACKING DEVICES w*W wx

Co

INTERACS DWATIBTASHE DHCIOSNATRICNASL MH)EIDSTCOARLY INTERACTSDWATIBTASHE DHCIOSANTRICANSL AH)CISTDOENRTY INTERACS DWAITBAHSE DHCIOSANTRICANSL E.G.
(
SAFETY
,
)
108
DATBSE 108a
STATS
F,
EG
(
ITNES
1080
SAFETY
,
EG
(
U.S. Patent Aug. 4 , 2020 Sheet 2 of 5 US 10,733,311 B2

202
SERVER PREDICTS USER BEHAVIOUR USING HISTORICAL DATA
204
DEVICE RECORDS INFORMATION ON CURRENT USER BEHAVIOUR
IS
CURRENT USER BEHAVIOUR DEVIATING FROM PREDICTED
NOJ TYPICAL BEHAVIOUR ? T YES
LOCAL
GATEWAY ANALYZES AND 208
REVIEWS DATA FROM DEVICE
m0owo0
GATEWAY SENDS REVIEWED AND COLLATED 210
DATA TO SERVER

PREMIUM
DETERMINATION SERVER INITIATES CALCULATION OF 212
UNIT INSURANCE PREMIUM BASED ON
224 CURRENT CIRCUMSTANCES
** * wwww W TO

SERVER MAINTAINS PREVIOUSLY PREMIUM = FN (EG .,WEATHER, **

214
DETERMINED INSURANCE PREMIUM SAFETY, ACCIDENT RATE , ***

TIME OF DAY,LOCATION )
SERVER SENDS UPDATED 216
INSURANCE PREMIUM TO
GATEWAY
LOCAL
GATEWAY SENDS REAL - TIME
ALERT AND PREMIUM UPDATES -218
TO USER'S DEVICE

CHANGES IN USER'S ACTIONS 220

RECORDED BY DEVICE AND
SENT TO GATEWAY
PROCESS IS REPEATED AFTER A DEFINED
DURATION OR IF CONTEXT OF USER CHANGES
222
FIG . 2
U.S. Patent Aug. 4 , 2020 Sheet 3 of 5 US 10,733,311 B2

USING HISTORICAL DATA, SERVER PREDICTS THAT USER WLL NOT BE HOME EVERYX NIGHT 302
OF THE WEEK (E.G.,WEEKEND NIGHT)
SMART METER DETECTS NETWORK ROUTER DETECTS USER'S CALENDAR DISPLAYS NO 304
ATYPICAL ELECTRICITY USAGE USER'S PHONE ACCESS POINT PLANNED SOCIAL EVENTS

IS 306
CURRENT USER BEHAVIOUR DEVIATING FROM PREDICTED
NO TYPICAL BEHAVIOUR ? HYES
???
LOCAL
GATEWAY ANALYZES AND REVIEWS 308
OCCUPANCY DATA ;AUGMENTS WITH KNOWN
NIGHT SAFETY IN NEIGHBOURHOOD

GATEWAY SENDS REVIEWED AND COLLATED 310

DATA TO INSURANCE COMPANY SERVER

PREMIUM 312
DETERMINATION SERVER INITIATES CALCULATION OF PREMIUM
UNT 322
3
BASED ON REAL -TIME OCCUPANCY DATA
$ w

SERVER MAINTAINS PREVIOUSLY

ex

1
w

PREMIUM = FN (E.G ., PAST SAFETY, 314

DETERMINED HOME INSURANCE TIME OF DAY, SUBURB , PROXIMITY 3
www

PREMIUM
3
h 6 4. V
TO KNOWN AREAS)
Man man m m MY Y Y Y Y Y
wir

mmmmmm wwwmmmmmm

SERVER SENDS USER'S 316

UPDATED HOME INSURANCE
PREMIUM TO GATEWAY

LOCAL
GATEWAY SENDS REAL - TIME 318
NOTIFICATION OF REDUCED
PREMIUM TO USER'S PHONE
?

PROCESS IS REPEATED AFTER A DEFINED

PERIOD (E.G.,AN HOUR )OR IFUSER'S PRESENCE 320
IS NO LONGER DETECTED
FIG . 3
U.S. Patent Aug. 4 , 2020 Sheet 4 of 5 US 10,733,311 B2

SERVER PREDICTS THAT USER WLL COMMUTE TO WORK VIA TRAIN IN CITY X ,COUNTRY Y 402

IN -CAR GPS RECORDS USER'S USER'S CALENDAR ENTRY INFORMS SERVER

DECISION TO DRIVE TO WORK IN COUNTRY ZITHAT USER IS CURRENTLY ABROAD IN COUNTRY Z
406
IS
CURRENT USER BEHAVIOUR DEVATING FROM PREDICTED
NO TYPICAL BEHAVIOUR ? LYES
LOCAL
GATEWAY ANALYZES AND REVIEWS USER'S 408
CHOSEN ROUTE AND DESTINATION ON GPS
GATEWAY SENDS REVIEWED USER LOCATION - 410
DATA TO SERVER
PREMIUM SERVER INITIATES CALCULATION OF
DETERMINATION INSURANCE PREMIUM BASED ON
UNIT 424 CURRENT CIRCUMSTANCES
4

SERVER MAINTAINS PREVIOUSLY PREMIUM = FN (E.G ., SAFETY AND ACCIDENTRATE ~ 414

DETERMINED INSURANCE E ama
ALONG ROUTE , TIME OF DAY. DESTINATION ,
?
PREMIUM taa
**

** INTENDED LOCATION OF PARKED VEHICLE )

SERVER SENDS UPDATED 416
INSURANCE PREMIUM TO
GATEWAY
LOCAL
GATEWAY SENDS REAL - TIME ALERT
AND PREMIUM UPDATES TO -418
USER'S DEVICE (EG .,GPS CAPABLE
D*EKR*E DEVICE ,PHONE)
CHANGES IN USER'S ACTIONS (E.G., 420
DIFFERENT ROUTE )RECORDED BY
DEVICE AND SENT TO GATEWAY
PROCESS IS REPEATED AFTER A DEFINED
DURATION OR IF CONTEXT OF USER CHANGES
422
FIG . 4
U.S. Patent Aug. 4 , 2020 Sheet 5 of 5 US 10,733,311 B2

PROCESSOR (S) MEMORY

16

30 STORAGE 18
GATEWAY SYSTEM

14 22
NETWORK 24
ADAPTOR
20 1/0
INTERFACE(S ) NETWORK
28
DISPLAY
26
DEVICE(S)

FIG . 5
 US 10,733,311 B2
1 2
COGNITIVE INTERNET OF THINGS (IOT) the sensor devices that are detecting user data in real- time.
GATEWAYS FOR DATA SECURITY AND The hardware processor may determine whether the real
PRIVACY PROTECTION IN REAL - TIME time data deviates from the predicted user behavior data .
CONTEXT-BASED DATA APPLICATIONS Responsive to determining that the real-time data deviates
5 from the predicted user behavior data , the hardware proces
FIELD sor may transform the real-time data into collated data that
removes protected and sensitive information from the real
The present application relates generally to computers and time data , and transmit the collated data to the server
computer applications , and more particularly to data security computer, wherein the protected and sensitive information
for real-time context-based data and cognitive computing. 10 received from the sensor devices remain local to the hard
ware processor. The hardware processor may receive from
BACKGROUND the server computer an updated offering, the updated offer
ing computed by the server computer based on the collated
High proliferation of wearable devices coupled with data. The hardware processor may transmit a real-time alert
uptake of Internet of Things ( IoT) devices provides data 15 to a user device to notify the user of the updated offering
connectivity as it has never been seen before. However, from the server computer.
using such data for real-time dynamic applications presents A computer readable storage medium storing a program
challenges in privacy protection and data security. of instructions executable by a machine to perform one or
Insurance for health , home, car and others is generally more methods described herein also may be provided .
computed in a relatively static manner. For instance , an 20 Further features as well as the structure and operation of
individual's health insurance may be computed based on various embodiments are described in detail below with
age, predisposition to illness, health habits , and others. reference to the accompanying drawings. In the drawings,
Home insurance is computed based on location of the home, like reference numbers indicate identical or functionally
when it was built , the type of construction materials used , similar elements .
and others . Car insurance is computed based on where it is 25
parked , how often it is driven ,how many drivers use the car, BRIEF DESCRIPTION OF THE DRAWINGS
the car make and model, and others. Once these factors are
determined , the insurance premium is computed only once , FIG . 1 is a diagram illustrating system architecture of a
and the amount is paid by the customer monthly or annually. system of the present disclosure in one embodiment.
These methods assume risks to be relatively static . 30 FIG . 2 is a flow diagram illustrating a method of the
present disclosure with reference to a general scenario .
BRIEF SUMMARY FIG . 3 is a flow diagram illustrating a method of the
present disclosure in one embodiment in an example sce
A system and method of providing a cognitive IoT nario .
gateway may be provided . The method may be executed by 35 FIG . 4 is a flow diagram illustrating a method in one
at least one hardware processor communicatively coupled to embodiment of the present disclosure with reference to
at least a server computer and sensor devices. The method, another example scenario .
in one aspect, may include receiving from the server com FIG . 5 illustrates a schematic of an example computer or
puter a predicted user behavior data associated with a user, processing system that may implement a cognitive IoT
wherein the server computer has provided an offering to the 40 gateway system in one embodiment of the present disclo
user based on the predicted user behavior. The method may sure .
also include receiving real- time data associated with the user
from the sensor devices that are detecting user data in DETAILED DESCRIPTION
real- time. The method may further include determining
whether the real- time data deviates from the predicted user 45 Cognitive gateways installed on local devices (e.g., a user
behavior data . The method may also include , responsive to device ) provide data security and privacy protection by
determining that the real - time data deviates from the pre leveraging data inputs available from various Internet of
dicted user behavior data , transforming the real-time data Things ( IoT) devices and /or sensors, smartphones, and other
into collated data that removes protected and sensitive such devices. A system and method in one embodimentmay
information from the real- time data , and transmitting the 50 leverage real -time dynamic data in a secure way, collected
collated data to the server computer, wherein the protected from these IoT sensors and /or devices , for an application in
and sensitive information received from the sensor devices insurance industry . For instance, the system and method may
remain local to the hardware processor. The method may determine insurance premium dynamically, for example , in
also include receiving from the server computer an updated real-time (e.g. , seconds, minutes or hours ), taking into
offering , the updated offering computed by the server com- 55 account that risks and circumstances surrounding the poli
puter based on the collated data . The method may further cyholders or entities are changing ( i.e., are dynamic ). The
include transmitting a real- time alert to a user device to system and method may leverage data in a secure manner
notify the user of the updated offering from the server from various Internet of Things (IoT) devices and /or sen
computer. sors, smartphones, and other such devices, via cognitive
A cognitive Internet of Things (IoT) gateway system , in 60 gateways, and take into account the context that the entity
one aspect, may include a hardware processor communica being insured (e.g., individuals , homes, cars , and others) is
tively coupled to a server computer and sensor devices. The currently in . The system and method may also provide
hardware processormay receive from the server computer a activity prediction in computing the pricing premium .
predicted user behavior data associated with a user,wherein The system and method of the present disclosure in one
the server computer has provided an offering to the user 65 embodiment may include local cognitive IoT gateways that
based on the predicted user behavior. The hardware proces can be used to determine premium in dynamic and real- time
sor may receive real-time data associated with the user from manner based on behavioral monitoring of IoT devices ,
 US 10,733,311 B2
3 4
while at the same time providing protection of privacy mobile phone notes that the location of the user has changed
and /or data security. The dynamic insurance pricing is from the typicalbase location (e.g., Country X ) to a different
cognizant of user's privacy and contextual information . location (e.g., Country Y ).
FIG . 1 is a diagram illustrating system architecture of a Examples of application context may include home and
system of the present disclosure in one embodiment. A 5 contents insurance. For example , an insurance company may
cognitive Internet of Things ( IoT) gateway 102 in one offer lower premium rates than normal when the presence of
embodiment monitors and stores (e.g., with the permission home owner is detected , for example, via phone connection
of users or individuals involved ) data associated with enti to home and/or WiFi network , which deviates from pre
ties, for example, vehicle, home, individual, and generates dicted home presence for particular time. Another example
reports autonomously and sends automatic reports of infor- 10 of application context may include travel and /or health
mation (absent any protected or private data ) that is needed insurance . Insurers may implement different premium rates
by a server 104 for an application , for example, responsive depending on the policyholder's location and mode of
to detecting changes in context surrounding the entity . The travel. For example , traveling in Country ABC versus trav
cognitive IoT gateway 102 generates the reports of infor eling in Country XYZ , walking in Location ABC versus
mation so that the information in the reports do not contain 15 taking a taxi versus using public transportation such as buses
any sensitive, protected and / or private data , yet the server or trains for commuting. Yet another application context
104 is able to use the information in its processing . For may include car insurance . Insurers may implement different
instance , any sensitive information remains in the cognitive premium rates if policyholder heeds travel advisory to use
IoT gateway 102 , local to that device, and is not transmitted different routes in lower risk areas or suburbs (via phone
to a server 104. The cognitive IoT gateway 102 may 20 GPS signal and historical area safety data ).
autonomously detect the context surrounding the entity Examples of protecting privacy in the application in
changes, generate a reportof information and send the report dynamic insurance pricing may include the following . For
of information to a server 104. An example of a cognitive example , a car insurance company may adjust the insurance
IoT gateway is a smartphone , a home modem , a wireless pricing based on Pay -as-you -drive (PAYD ) and Pay -how
router, a single board computer , and other devices, having at 25 you -drive (PHYD ). One factor that may change the price in
least a hardware processor and software components , and PHYD is road types (paved or unpaved ). In this example
communication functionalities. The cognitive IoT gateway case , the insurance company may need to know the car
is coupled to sensor and/or other IoT devices 106 that can location to determine the road type of the road that an
collect data or information from the entities. The cognitive insured is driving on . In the present disclosure in one
IoT gateway 102 may be preloaded by apps (e.g., smart- 30 embodiment, instead of directly communicating the specific
phone application ) from the server 104 , for example , for or exact location of the car (and therefore the driver or the
communicating data or information to and from the server insured ), the IoT gateway with a preloaded map and a local
104 for its application . An example application is in the GPS and , for example , other IoT sensors can examine the
insurance industry . The server 104 may be associated with road type according to the location of the car and send a
an insurance company , for example, an insurance company's 35 report about the percentage of time that the car spent driving
server that functions to support the insurance company. An on particular road type, for example, unpaved or paved road .
app provided by the insurance company's server may be In one embodiment, an insurance company's server may
preloaded onto a smartphone device that functions as a determine the insurance premium predictively based on
cognitive IoT gateway. For example , a cognitive IoT gate historical travel patterns and anticipated routes taken by the
way can be pre - loaded by one or more apps from an 40 user. The insurance company's server may determine a
insurance company for the purposes of communicating driving profile of an insured , for example , using historical
information to the insurance servers, communicating data, and determine the premium rate based on the driving
dynamic pricing to an individual, and to ascertain any profile. Thereafter , the insurance company's server may
changes to the contextual information. determine incremental changes to the premium rate based on
Examples of sensor and IoT devices 106 may include, but 45 the dynamic information received from a cognitive IoT
are not limited to wearable devices, location tracking gateway. The information that is received from the cognitive
devices such as global positioning system (GPS ) devices, IoT gateway does not contain any protected or private
cameras installed in the areas such as homes,motion detec information concerning the insured , thereby preserving data
tors, utility usage sensors or readers , other devices thatmay privacy.
detect occupancy and /or location . The sensor and IoT 50 Referring to FIG . 1, sensors and IoT devices 106 com
devices 106 may continuously detect current context of municate or interact with databases 108. A database at 108a
entities and send real-time data to the cognitive IoT gateway may store historical data , for example , fitness statistics of an
102. In one embodiment, each of the sensor and IoT devices individual, medical history of an individual, and /or other
106 sends its real-time data to the cognitive IoT gateway. health information. Health information pertaining to a par
The context of the entities is deemed to have changed by the 55 ticular individual such as number of steps taken daily and
cognitive IoT gateway when the current real-time data heart rate are received from IoT devices 106 such as fitness
deviates from predicted and /or past user behavior. Predicted trackers and smartphones equipped with a pedometer , and
user behavior is computed via application ofmachine learn can be used to provide a more accurate health profile of an
ing techniques such as regression and neural network algo individual for premium pricing purposes. A database at 1086
rithms on historical data that may be stored on the IoT 60 may store historical data including safety information about
devices or in a database , while past user behavior is deter an area or neighborhood , accident history of an area or
mined from data stored in the sensor and/or IoT devices neighborhood , road information about an area, and other
which have built - in memory capabilities . Change of context driving or parking conditions about an area or neighbor
can be inferred by the cognitive IoT gateway when for hood,map of an area , and/or other data about an area . These
example the home IoT gateway determines that the users' 65 types of information are received from IoT devices 106
devices are not connecting to it over the timeframe when capable of geography tracking, such as GPS units and
they normally establish a connection . Or when the users ' smartphones. For example , GPS units are preloaded with
 US 10,733,311 B2
5 6
road information and other parking conditions while smart insurance server. In another example, an up -to -date general
phones provide up - to -date information about safety alerts at health profile of an individual ( e.g., duration of high inten
an insured's current location . A database at 108c may store sity activity ) can be determined from information collected
historical data that may include safety information associ by and received from fitness trackers ( e.g., times of day
ated with an area or neighborhood and other risk information 5 when the individual walks or is active, number of stairs
associated with an area or neighborhood . For example , the climbed ,heart rate ). An anomaly detection technique such as
information can be received from electronic communication cluster analysis can be used to ascertain that the insured is
on smartphone devices 106 and /or information from local exercising more than normal (based on historical health
government or like websites . data ), in which case the cognitive IoT gateway sends the
Sensors and IoT devices 106 detect real-time data , for 10 duration of high intensity activity to the insurance server.
example, current location , whether a building or home is A server at 104 stores, for example, data relevant to
currently occupied . In one embodiment, a home may be insurance premium pricing and determines insurance pre
determined to be occupied when the home Wi-Fi router / mium prices. The server 104 receives the generic informa
access point ( i.e. cognitive IoT gateway ) detects the insured tion generated by the cognitive IoT gateway 102 and based
individual's mobile phone/ laptop/ tablet, or another such 15 on the information computes a dynamic premium . For
device . The location of the individual can also be determined instance , for determining a dynamic premium on car insur
through calendar entries in the individual's smartphone ance , if the information indicates that the insured is spending
which usually contains venue information of an event the more time in driving on roads that have different conditions
individual may be attending . In another example, informa than before, the premium may be dynamically adjusted to
tion on electricity usage in an insured's home, collected via 20 reflect that information . The server 104 sends updated
smart meters, can indicate the home occupancy status . These insurance premium pricing to the cognitive IoT gateway
IoT devices then send the data to a cognitive IoT gateway 102. The cognitive IoT gateway 102 sends a real- time alert
102 at every specified interval. The interval may be config to a user's device (e.g. , smartphone preloaded with the
ured , for example, predefined . insurance company's app ) notifying the user of the adjusted
The cognitive IoT gateway 102 receives, reviews and 25 premium . For example , an insured's home router/single
analyzes the data , and determines if the context of the board computer device /smartphone ( for example, acting as
insured has changed . For example , when insured individual the cognitive IoT gateway ) receives a push notification from
moves out of the individual's base location or when there are the insurance company server of the change in premium
no occupants in the house. The gateway generates a report price . The user's cognitive IoT gateway then sends an alert
that only includes information needed by the server 104 for 30 to the insured , such as a text message, or push notification ,
its application, stripping out any private or protected aspect (shown on the smartphone, or on one or more wearable
of the data . For instance , the cognitive IoT gateway 102may devices) notifying the insured of the change in premium
transform or convert a specific location data into a generic price.
form that does not reveal the specific location . For instance , FIG . 2 is a flow diagram illustrating a method in one
the cognitive IoT gateway 102 may determine the type of 35 embodiment of the present disclosure with reference to a
information needed by the server 104 and determine what general scenario . At 202 , a server predicts user behavior
type the specific location data falls into . The generic form of using historical data . A known prediction algorithm such as
data that is generated from the specific location data may be regression or neural network algorithm may be employed to
the amount of time an insured has spent in that type of predict user behavior based on historical data .
location . For example , the cognitive IoT gateway 102 may 40 At 204 , one or more sensor devices or IoT devices record
determine based on the specific location , the condition of the information on current user behavior. For example , one or
roads at that location , and only send to the server, the amount more fitness trackers with in -built memory record informa
of time that is spent on routes having that condition . To tion such as the number of steps the user has taken in a day,
illustrate , the cognitive IoT gateway, having received infor heart rate , the duration the user exercises , and send the
mation on current road conditions from geography tracking 45 information to the database 108a . These wearable devices
devices, reviews the information and compares it with the and /or sensor devices send their real-time data to the cog
insured's historical travel data (stored in database 108b ). In nitive IoT gateway, which could be the users ' smartphone .
one embodiment, one or more machine learning techniques The context of the entities is deemed to have changed when
can be used to determine if the road conditions of the the current real-time data deviates from predicted and /or
insured's current location are different from the insured's 50 past user behavior. Predicted user behavior is computed via
typical road conditions (e.g., rough terrains versus paved application of machine learning techniques such as regres
highway roads ). In such situations, the cognitive IoT gate sion and neural network algorithms on historical data that
way sends the duration of time spent on the current road may be stored on the IoT devices or in a database , while past
conditions to the insurance company server. As another user behavior is determined from data stored in the sensor
example , information such as whether a home or another 55 and /or IoT devices which have built- in memory capabilities .
building is currently occupied may be transformed to the At 206 , the cognitive IoT gateway determines whether the
amount of time that home or building has occupancy. In this current user behavior is deviating from the predicted behav
scenario , for example , the cognitive IoT gateway receives ior and sends the report to the server. The cognitive IoT
information from the insured's home router that the insured gateway performs the check in order to protect users '
is at home on a particular time, for example, day of theweek , 60 privacy. For example, the data predicted at 202 is compared
and compares this information with historical data stored in with the data of the current user behavior. The server
database 1080 (assume for example , the insured is usually predicts the best behavior/action to users according to his
not home on this day of the week ). In one embodiment, the torical data from the insurance database and recommenda
cognitive IoT gateway uses an anomaly detection technique tions (e.g., road safety, hazard ) and sends the predicated
such as k -nearest neighbor algorithm to determine that this 65 route to the IoT gateway to inform the user. Then , the actual
behavior deviates from the insured's normal routine and action /behavior is identified using embedded sensors and
sends the duration of time the insured is home to the IoT devices ( e.g., GPS , movement trackers ) connected to the
 US 10,733,311 B2
7 8
gateway. The cognitive IoT gateway compares the predi access point. Yet as another example, user's electronic
cated action /behavior with the current actual action / behav calendar stores or displays no planned social events .
ior. At 306 , the cognitive IoT gateway determines whether the
The current user behavior is determined to be deviating current user behavior is deviating from the predicted behav
from the predicted behavior if the difference between the 5 ior, using a machine learning technique such as a cluster
two data is greater that a threshold value or outside of a analysis and a k -nearest neighbors algorithm . For example ,
marginal value. In some embodiments, this can be deter the data predicted at 302 is compared with the data of the
mined using anomaly detection techniques such as cluster current user behavior determined at 304. For example, the
analysis and k -nearest neighbors . server sends the predicted behavior to the cognitive IoT
If the current user behavior is deviating from the predicted 10 gateway.
user behavior, at 208 , a gateway device that is local to the The current user behavior is determined to be deviating
user analyzes and reviews data from the one or more sensor from the predicted behavior if the difference between the
devices. The gateway device then generates a collated data two data are greater that a threshold value or outside of a
that does not include private or protected information . At marginal value.
210 , the gateway sends the collated data to the server. In this 15 If the currentuser behavior is deviating from the predicted
way, private and protected information stays within the user behavior, at 308, a gateway device that is local to the
locality of the user, e.g., user's own device , and is not user analyzes and reviews occupancy data from the one or
disseminated to another recipient device such as a compa more sensor devices and augments the information with
ny's server. known safety information about the neighborhood . The
At 212 , the server (e.g., server computer ) receives the 20 gateway device then generates collated data that does not
collated data, and initiates computation of insurance pre include private or protected information . For example , the
mium based on current circumstance indicated by the col gateway, having received information about user's (lack of)
lated data . calendar entry on a Saturday night from user's smartphone
For example , as shown at 214 , a premium may be and /or higher electricity usage from user's smart meter,
computed as a function of weather, safety information , 25 sends information on the duration for which the user is
accident rate and time ofday. The gateway is preloaded with home, as opposed to sending the currentwhereabouts of the
a map from insurance company. The insurance company can user. At 310 , the gateway sends the collated data to the
update the map over time by sending information such as server. In this way, private and protected information stays
safety, accident rate, etc. to give the best premium . within the locality of the user, e.g., user's own device , and
At 216 , the server sends the updated insurance premium 30 is not disseminated to another device such as a company's
to the cognitive IoT gateway . server .
At 218, the cognitive IoT gateway receives the updated At 312, the server (e.g., server computer ) receives the
insurance premium and sends a real-time alert and premium collated ta , and initiates computation of insurance pre
update to a user's device. mium based on real- time occupancy data indicated by the
At 220 , sensor and /or IoT devices continue to detect and 35 collated data . For example, as shown at 314, a base premium
record user actions, and changes are recorded and sent to the may be computed as a function of safety information of the
gateway device . At 222 , the process repeats after a config area or neighborhood , time of day, suburb , and proximity to
ured duration or if the context of the user changes . An law enforcement station (e.g., police station ). The collated
example of context change in user is the user driving on occupancy information provided by the cognitive IoT gate
paved roads again ( that is, the user's normal routine) after 40 way provides additional input to the computation and is used
being on rough terrains for a definite amount of time. In to augment the base premium . The updated home insurance
another example, the context of a user is deemed to have premium may be adjusted from the insured's base premium
changed when the user who is home on a Saturday evening if, for example, the insured is not home during the times that
( contrary to the user's historical data which indicates the the user's is normally home. The final (updated ) home
user is usually not home on Saturday night) goes out later 45 insurance premium is the premium that is communicated to
that night. For instance , after a configured duration expires the insured .
and /or responsive to detecting a change in user context, the At 316 , the server sends the updated home insurance
logic of the process returns to 206 to iterate the processing . premium to the cognitive IoT gateway .
At 206 , responsive to detecting that there is no deviation At 318, the cognitive IoT gateway receives the updated
or the deviation is within a threshold , the processing may 50 home insurance premium and sends a real -time notification
repeat, for example, after an expiration a configured duration of the homepremium update to a user's device , for example ,
or responsive to detecting a changing in user context at 222 . a user's phone.
In this scenario , the server maintains previously determined At 320 , the process repeats after a configured duration of
insurance premium as shown at 224 . time or if the context of the user changes, for example, the
FIG . 3 is a flow diagram illustrating a method of the 55 user's presence is no longer detected . For instance , after a
present disclosure in one embodiment in an example sce configured duration expires and /or responsive to detecting
nario , for example, in determining home insurance premium absence of the user at home, the logic of the process returns
based on dynamic context based data . At 302 , a server to 306 to iterate the processing.
predicts user behavior using historical data. In this example , At 306 , responsive to detecting that there is no deviation
the server predicts that a user will not be home every 60 or the deviation is within a threshold , the processing may
Saturday night. A known prediction algorithm such as repeat, for example , after an expiration of a configured
regression or neural network algorithm may be employed to duration or responsive to detecting user absence at 320. In
predict user behavior based on historical data . this scenario , the server maintains the previously determined
At 304 , one or more sensor devices or IoT devices record home insurance premium as shown at 322 .
information on current user behavior. For example, a smart 65 FIG . 4 is a flow diagram illustrating a method in one
meter detects atypical electricity (or another utility ) usage . embodiment of the present disclosure with reference to an
As another example, a network router detects user's phone example scenario . At 402 , a server predicts user behavior
 US 10,733,311 B2
9 10
using historical data . In this example , the server predicts that gateway device. In this example, different user routes may
the user will commute to work via public transportation , for be detected . At 422 , the process repeats after a configured
example , a train in City A , in Country Z. A known prediction duration or if the context of the user changes . An example
algorithm such as regression or neural network algorithm of context change in user is the user driving on paved roads
may be employed to predict user behavior based on histori- 5 again (that is , the user's usual routine ) after being on rough
cal data . terrains for a definite amount of time. Another example of
At 404 , one or more sensor devices or IoT devices detect context change is the user returning to the user's country of
and record information on current user behavior. In one residence. For instance, after a configured duration expires
aspect, sensors have memory that enables the sensors to and /or responsive to detecting a change in user context, the
record user behavior. For example, an in -car GPS device 10 logic of the process returns to 406 to iterate the processing .
records user's decision to drive to work in Country X , and At 406 , responsive to detecting that there is no deviation
the user's electronic calendar entry indicates and notifies a or the deviation is within a threshold , the processing may
cognitive IoT gateway, which further notifies a server that repeat, for example , after an expiration of a configured
the user is currently abroad in Country X. duration or responsive to detecting a change in the user's
At 406 , a cognitive IoT gateway determines whether the 15 route at 422. In this scenario , the server maintains the
current user behavior is deviating from the predicted (or previously determined home insurance premium as shown at
recommended ) behavior. For example , the data predicted at 424 .
402 is compared with the data of the current user behavior. The user has a preloaded map (or another application
The server may recommend the best (suitable ) behavior/ from the insurance company) locally in the user's cognitive
action to the users based on the policies and rules from the 20 IoT gateway ( for example , a mobile phone ). The location of
insurance company . For example , the recommended behav the user can be identified on the map and the user can enter
ior/ action may be sent tothe IoT gateways, which sends the the destination on the map ( all information are local) and the
recommendation to the user. Then the IoT gateway can map which is powered by insurance company's rules and
identify the user behavior/action per the data received from policies can offer the best solution (safest) for that specific
IoT devices and sensors. The IoT gateway can compare the 25 area to the user. The user preferences can also be considered ,
recommended behavior/action to the actual user behavior/ for example, whether the user prefers to use a car or public
action and send a report to the server. transport. After recommending an action /behavior, the cog
The current user behavior is determined to be deviating nitive IoT gateway can identify whether or not the user
from the predicted behavior if the difference between the followed the recommended action /behavior ( for example ,
two data is greater that a threshold value or outside of a 30 whether or not the user traveled on the recommended route ).
marginal value . In some embodiments , this can be deter If it is determined that the user did not travel on the
mined using one or more anomaly detection techniques such recommended rout, the actual route can be identified and the
as cluster analysis and k -nearest neighbors algorithm . cognitive IoT gateway may send a report to the server, which
If the current user behavior is deviating from the predicted may recompute the premium .
user behavior, at 408 , a gateway device that is local to the 35 FIG . 5 illustrates a schematic of an example computer or
user analyzes and reviews data from the one or more sensor processing system that may implement a cognitive IoT
devices. In this example , the gateway device analyzes and gateway system in one embodiment of the present disclo
reviews the user's chosen route and destination on theGPS . sure . The computer system is only one example of a suitable
The gateway device then generates a collated data that does processing system and is not intended to suggest any limi
not include private or protected information . At 410 , the 40 tation as to the scope of use or functionality of embodiments
gateway sends the collated data to the server. In this way, of the methodology described herein. The processing system
private and protected information stays within the locality of shown may be operational with numerous other general
the user, e.g., user's own device, and is not disseminated to purpose or special purpose computing system environments
another recipient device such as a company's server. or configurations. Examples of well-known computing sys
At 412, the server (e.g., server computer ) receives the 45 tems, environments, and/or configurations that may be suit
collated data , and initiates computation of insurance pre able for use with the processing system shown in FIG . 5 may
mium based on current circumstance indicated by the col include, but are not limited to , personal computer systems,
lated data . For example, as shown at 414 , a premium may be server computer systems, thin clients , thick clients, handheld
computed as a function of safety factors of the area , accident or laptop devices, multiprocessor systems, microprocessor
rate along the current route, time of day , destination , 50 based systems, set top boxes, programmable consumer elec
intended location of parked vehicle . The information such as tronics, network PCs , minicomputer systems, mainframe
the location of the user, the destination or the exact route that computer systems, and distributed cloud computing envi
the user decides to take are examples of information not sent ronments that include any of the above systems or devices ,
to the server . Information that does not protect the privacy and the like .
of the individual is not sent to the server . 55 The computer system may be described in the general
At 416 , the server sends the updated insurance premium context of computer system executable instructions, such as
to the cognitive IoT gateway . program modules, being executed by a computer system .
At 418 , the cognitive IoT gateway receives the updated Generally , program modules may include routines, pro
insurance premium and sends a real- time alert and premium grams, objects , components , logic , data structures, and so on
update to a user's device, for example a user's phone or 60 that perform particular tasks or implement particular abstract
other GPS - enabled devices which may include fitness track data types. The computer system may be practiced in
ers and smart watches. The real-time alert and premium distributed cloud computing environments where tasks are
update may be in the form of a text message to a smart performed by remote processing devices that are linked
phone, or push notifications on one or more wearable through a communications network . In a distributed cloud
devices . 65 computing environment, program modules may be located
At 420 , sensor and/ or IoT devices continue to detect and in both local and remote computer system storage media
record user actions, and changes are recorded and sent to the including memory storage devices .
 US 10,733,311 B2
11 12
The components of computer system may include, but are The computer readable storage medium can be a tangible
not limited to , one ormore processors or processing units 12 , device that can retain and store instructions for use by an
a system memory 16 , and a bus 14 that couples various instruction execution device . The computer readable storage
system components including system memory 16 to proces medium may be , for example, but is not limited to , an
sor 12. The processor 12 may include a cognitive IoT 5 electronic storage device, a magnetic storage device, an
gateway module 30 that performs the methods described optical storage device, an electromagnetic storage device, a
herein . The module 30 may be programmed into the inte of semiconductor storage device , or any suitable combination
grated circuits of the processor 12 , or loaded from memory examples the foregoing. A non -exhaustive list of more specific
16 , storage device 18 , or network 24 or combinations 10 the following of the computer readable storagemedium includes
thereof. : a portable computer diskette , a hard disk , a
Bus 14 may represent one or more of any of several types random access memory (RAM ), a read -only memory
of bus structures, including a memory bus or memory ((ROM ), an erasable programmable read -only memory
controller , a peripheral bus, an accelerated graphics port, and (EPROM or Flash memory ), a static random access memory
SRAM ), a portable compact disc read -only memory (CD
aarchitectures
processor .orBylocal bus using any of a variety of bus 15 ROM ), a digital versatile disk (DVD ), a memory stick, a
way of example, and not limitation , such floppy disk , a mechanically encoded device such as punch
architectures include Industry Standard Architecture (ISA ) cards or raised structures in a groove having instructions
bus,Micro Channel Architecture (MCA )bus , Enhanced ISA recorded thereon , and any suitable combination of the fore
(EISA ) bus , Video Electronics Standards Association going. A computer readable storage medium , as used herein ,
(VESA ) local bus , and Peripheral Component Interconnects 20 is not to be construed as being transitory signals per se , such
(PCI) bus. as radio waves or other freely propagating electromagnetic
Computer system may include variety of computer waves, electromagnetic waves propagating through a wave
system readable media. Such media may be any available guide or other transmission media (e.g., light pulses passing
media that is accessible by computer system , and it may through a fiber-optic cable ), or electrical signals transmitted
include both volatile and non -volatile media, removable and 25 through a wire .
non -removable media . Computer readable program instructions described herein
System memory 16 can include computer system readable can be downloaded to respective computing/processing
media in the form of volatile memory , such as random devices from a computer readable storage medium or to an
access memory (RAM ) and/ or cache memory or others. external computer or external storage device via a network ,
Computer system may further include other removable /non- 30 for example , the Internet, a local area network , a wide area
removable, volatile/non -volatile computer system storage network and /or a wireless network . The network may com
media . By way of example only, storage system 18 can be prise copper transmission cables, optical transmission fibers,
provided for reading from and writing to a non -removable , wireless transmission , routers, firewalls, switches , gateway
non -volatile magnetic media (e.g., a “ hard drive” ). Although computers and/or edge servers . A network adapter card or
not shown, a magnetic disk drive for reading from and 35 network interface in each computing processing device
writing to a removable, non - volatile magnetic disk (e.g., a receives computer readable program instructions from the
“ floppy disk ” ), and an optical disk drive for reading from or network and forwards the computer readable program
writing to a removable, non -volatile optical disk such as a instructions for storage in a computer readable storage
CD -ROM , DVD -ROM or other optical media can be pro medium within the respective computing/processing device .
vided . In such instances , each can be connected to bus 14 by 40 Computer readable program instructions for carrying out
one or more data media interfaces. operations of the present invention may be assembler
Computer system may also communicate with one or instructions, instruction -set-architecture (ISA ) instructions ,
more external devices 26 such as a keyboard, a pointing machine instructions, machine dependent instructions,
device, a display 28 , etc .; one or more devices that enable a microcode, firmware instructions, state -setting data , or
user to interact with computer system ; and /or any devices 45 either source code or object code written in any combination
(e.g., network card , modem , etc.) that enable computer ofone or more programming languages , including an object
system to communicate with one or more other computing oriented programming language such as Smalltalk , C ++ or
devices. Such communication can occur via Input/Output the like, and conventional procedural programming lan
(1/0 ) interfaces 20. guages, such as the “ C ” programming language or similar
Still yet , computer system can communicate with one or 50 programming languages . The computer readable program
more networks 24 such as a local area network (LAN ) , a instructions may execute entirely on the user's computer,
general wide area network (WAN ), and /or a public network partly on the user's computer, as a stand -alone software
(e.g., the Internet) via network adapter 22. As depicted , package , partly on the user's computer and partly on a
network adapter 22 communicates with the other compo remote computer or entirely on the remote computer or
nents of computer system via bus 14. It should be under- 55 server. In the latter scenario , the remote computer may be
stood that although not shown, other hardware and/or soft connected to the user's computer through any type of
ware components could be used in conjunction with network , including a local area network (LAN ) or a wide
computer system . Examples include , but are not limited to : area network (WAN ), or the connection may be made to an
microcode, device drivers, redundant processing units , external computer ( for example , through the Internet using
external disk drive arrays, RAID systems, tape drives, and 60 an Internet Service Provider ). In some embodiments , elec
data archival storage systems, etc. tronic circuitry including, for example ,programmable logic
The present invention may be a system , a method, and /or circuitry, field -programmable gate arrays (FPGA ), or pro
a computer program product. The computer program prod grammable logic arrays (PLA ) may execute the computer
uct may include a computer readable storage medium (or readable program instructions by utilizing state information
media ) having computer readable program instructions 65 of the computer readable program instructions to personalize
thereon for causing a processor to carry out aspects of the the electronic circuitry, in order to perform aspects of the
present invention . present invention.
 US 10,733,311 B2
13 14
Aspects of the present invention are described herein with the claims below are intended to include any structure ,
reference to flowchart illustrations and /or block diagrams of material, or act for performing the function in combination
methods , apparatus (systems), and computer program prod with other claimed elements as specifically claimed . The
ucts according to embodiments of the invention . It will be description of the present invention has been presented for
understood that each block of the flowchart illustrations 5 purposes of illustration and description , but is not intended
and / or block diagrams, and combinations of blocks in the to be exhaustive or limited to the invention in the form
flowchart illustrations and/or block diagrams, can be imple disclosed.Many modifications and variations will be appar
mented by computer readable program instructions. ent to those of ordinary skill in the art without departing
These computer readable program instructions may be from the scope and spirit of the invention . The embodiment
provided to a processor of a general purpose computer, 10 was chosen and described in order to best explain the
special purpose computer, or other programmable data pro principles of the invention and the practical application , and
cessing apparatus to produce a machine, such that the to enable others of ordinary skill in the art to understand the
instructions, which execute via the processor of the com invention for various embodiments with various modifica
puter or other programmable data processing apparatus, tions as are suited to the particular use contemplated .
create means for implementing the functions/acts specified 15 We claim :
in the flowchart and /or block diagram block or blocks. These 1. A cognitive Internet of Things (IoT) gateway system ,
computer readable program instructions may also be stored comprising
in a computer readable storage medium that can direct a a hardware processor communicatively coupled to a
computer, a programmable data processing apparatus, and / server computer and sensor devices ;
or other devices to function in a particular manner, such that 20 the hardware processor receiving from the server com
the computer readable storage medium having instructions puter a predicted user behavior data associated with a
stored therein comprises an article of manufacture including user, wherein the server computer has provided an
instructions which implement aspects of the function /act offering to the user based on the predicted user behav
specified in the flowchart and /or block diagram block or ior, the predicted user behavior data computed by
blocks. 25 executing a machine learning algorithm on historical
The computer readable program instructions may also be data stored in a database ;
loaded onto a computer, other programmable data process the hardware processor receiving real -time data associ
ing apparatus, or other device to cause a series of operational ated with the user from the sensor devices that are
steps to be performed on the computer, other programmable detecting user data in real- time,
apparatus or other device to produce a computer imple- 30 the hardware processor determining whether the real- time
mented process, such that the instructions which execute on data deviates from the predicted user behavior data, the
the computer, other programmable apparatus, or other predicted userbehavior data being data based on which
device implement the functions/acts specified in the flow the offering was provided ,
chart and/ or block diagram block or blocks . responsive to determining that the real -time data deviates
The flowchart and block diagrams in the Figures illustrate 35 from the predicted user behavior data , the hardware
the architecture, functionality , and operation of possible processor transforming the real-time data into collated
implementations of systems, methods, and computer pro data that removes protected and sensitive information
gram products according to various embodiments of the from the real-time data, and transmitting the collated
present invention . In this regard , each block in the flowchart data to the server computer, wherein the protected and
or block diagrams may represent a module , segment, or 40 sensitive information received from the sensor devices
portion of instructions, which comprises one or more remain local to the hardware processor,
executable instructions for implementing the specified logi the hardware processor receiving from the server com
cal function (s). In some alternative implementations, the puter an updated offering , the updated offering com
functions noted in the block may occur outof the order noted puted by the server computer based on the collated
in the figures. For example , two blocks shown in succession 45 data , and
may, in fact , be executed substantially concurrently , or the the hardware processor transmitting a real-time alert to a
blocks may sometimes be executed in the reverse order, user device to notify the user of the updated offering
depending upon the functionality involved . It will also be from the server computer,
noted that each block of the block diagrams and/or flowchart wherein the hardware processor transforming the real
illustration , and combinations of blocks in the block dia- 50 time data into collated data that removes protected and
grams and /or flowchart illustration , can be implemented by sensitive information from the real-time data comprises
special purpose hardware -based systems that perform the at least the hardware processor transforming a geo
specified functions or acts or carry out combinations of graphical coordinate location data into an amount of
special purpose hardware and computer instructions . time spent on a type of location the geographical
The terminology used herein is for the purpose of describ- 55 coordinate location data can be categorized into , the
ing particular embodiments only and is not intended to be hardware processor transmitting the collated data with
limiting of the invention . As used herein , the singular forms out the geographical coordinate location data ,
“ a” , “ an ” and “ the ” are intended to include the plural forms wherein the offering is an insurance premium and the
as well ,unless the context clearly indicates otherwise . It will server computer is associated with an insurance indus
be further understood that the terms “ comprises” and /or 60 try providing insurance to the user, wherein the insur
" comprising," when used in this specification , specify the ance premium that is offered to the user is dynamically
presence of stated features, integers , steps, operations, ele adjusted based on user's real- time behavior detected by
ments, and /or components, but do not preclude the presence the sensor devices ,
or addition of one or more other features, integers, steps, wherein the offering is a home insurance premium and the
operations , elements , components, and / or groups thereof. 65 user's real- time behavior comprises occupancy of the
The corresponding structures, materials, acts , and equiva user's home, wherein the hardware processor deter
lents of all means or step plus function elements, if any, in mines whether the real-time data deviates from the
 US 10,733,311 B2
15 16
predicted user behavior data by detecting that connec 7. Themethod of claim 6 , wherein the hardware processor
tion is not established between the hardware processor is a component of a smartphone associated with the user.
and the user device during a specified timeframe. 8. The method of claim 6 , wherein the sensors comprise
2. The system of claim 1, wherein the hardware processor one or more ofa wearable device, a location tracking device ,
is a component of a smartphone associated with the user. 5 a camera, motion detector, and a utility usage sensor.
3. The system of claim 1 , wherein the sensors comprise 9. The method of claim 6 , wherein the hardware processor
one or more of a wearable device, a location tracking device , is deployed with an app received from the server computer
a camera , motion detector, and a utility usage sensor. and the hardware processor performs the receiving steps, the
4. The system of claim 1 , wherein the hardware processor determining steps, the transforming, and the transmitting
is deployed with an app received from the server computer 10 steps by executing the app .
and the hardware processor performs the receiving steps , the 10. The method of claim 6 , wherein the offering further
determining steps, the transforming, and the transmitting includes a car insurance premium and the user's real- time
steps by executing the app. behavior comprises road conditions of a route in which the
5. The system of claim 1, wherein the offering further user is currently driving .
includes a car insurance premium and the user's real- time 15 11. A computer program product for providing a cognitive
behavior comprises road conditions of a route in which the IoT gateway, the computer program product comprising a
user is currently driving . computer readable storage medium having program instruc
6. A method of providing a cognitive IoT gateway, the tions embodied therewith , the program instructions execut
method executed by at least one hardware processor com able by a device to cause the device to :
municatively coupled to at least a server computer and 20 receive from a server computer a predicted user behavior
sensor devices, comprising: data associated with a user, wherein the server com
receiving from the server computer a predicted user puter has provided an offering to the user based on the
behavior data associated with a user, wherein the server predicted user behavior, the predicted user behavior
computer has provided an offering to the user based on data computed by executing a machine learning algo
the predicted user behavior, the predicted user behavior 25 rithm on historical data stored in a database ;
data computed by executing a machine learning algo receive real- time data associated with the user from
rithm on historical data stored in a database ; sensor devices that are detecting user data in real-time;
receiving real-time data associated with the user from the determine whether the real-time data deviates from the
sensor devices that are detecting user data in real-time; predicted user behavior data , the predicted user behav
determining whether the real-time data deviates from the 30 ior data being data based on which the offering was
predicted user behavior data , the predicted user behav provided ;
ior data being data based on which the offering was responsive to determining that the real- time data deviates
provided ; from the predicted user behavior ata , transform the
responsive to determining that the real-time data deviates real -time data into collated data that removes protected
from the predicted user behavior data , transforming the 35 and sensitive information from the real-time data , and
real- time data into collated data that removes protected transmit the collated data to the server computer,
and sensitive information from the real- time data, and wherein the protected and sensitive information
transmitting the collated data to the server computer, received from the sensor devices remain local to the
wherein the protected and sensitive information machine ;
received from the sensor devices remain local to the 40 receive from the server computer an updated offering, the
hardware processor; updated offering computed by the server computer
receiving from the server computer an updated offering , based on the collated data ; and
the updated offering computed by the server computer transmit a real-time alert to a user device to notify the user
based on the collated data ; and of the updated offering from the server computer,
transmitting a real-time alert to a user device to notify the 45 wherein the offering is an insurance premium and the
user of the updated offering from the server computer, server computer is associated with an insurance indus
wherein the transforming the real-time data into collated try providing insurance to the user, wherein the insur
data that removes protected and sensitive information ance premium that is offered to the user is dynamically
from the real-time data comprises at least transforming adjusted based on user's real-timebehavior detected by
a geographical coordinate location data into an amount 50 the sensor devices,
of time spent on a type of location the geographical wherein the offering is a home insurance premium and the
coordinate location data can be categorized into , user's real- time behavior comprises occupancy of the
wherein the collated data is transmitted to the server user's home, wherein the device is caused to determine
without the geographical coordinate location data , whether the real- time data deviates from the predicted
wherein the offering is an insurance premium and the 55 user behavior data by detecting that connection is not
server computer is associated with an insurance indus established between the device and the user device
try providing insurance to the user, wherein the insur during a specified timeframe.
ance premium that is offered to the user is dynamically 12. The computer program product of claim 11 , wherein
adjusted based on user's real-time behavior detected by the machine is a componentof a smartphone associated with
the sensor devices , 60 the user.
wherein the offering is a home insurance premium and the 13. The computer program product of claim 11, wherein
user's real- time behavior comprises occupancy of the the sensors comprise one or more of a wearable device, a
user's home, wherein the hardware processor deter location tracking device , a camera, motion detector, and a
mines whether the real- time data deviates from the utility usage sensor.
predicted user behavior data by detecting that connec- 65 14. The system of claim 1 , wherein the hardware proces
tion is not established between the hardware processor sor transforming a geographical coordinate location data
and the user device during a specified timeframe. into an amount of time spent on a type of location the
 US 10,733,311 B2
17 18
geographical coordinate location data can be categorized
into , comprises at least ,the hardware processor transforming
the geographical coordinate location data into the amount of
time spent on an unpaved road .
15. The method of claim 6 , wherein the transforming a 5
geographical coordinate location data into an amountof time
spent on a type of location the geographical coordinate
location data can be categorized into , comprises at least,
transforming the geographical coordinate location data into
the amount of time spent on an unpaved road . 10