`Case 6:20-cv-01063-ADA Document 1-1 Filed 11/17/20 Page 1 of 16
`
`EXHIBIT A
`
`EXHIBIT A
`
`
`
`Case 6:20-cv-01063-ADA Document 1-1 Filed 11/17/20 Page 2 of 16
`Case 6:20'CV'01063'ADA D"C“ml‘lllll11||1||llflllll11111111111111111111111111111||||||||
`
`US010142791B2
`
`(12) United States Patent
`US 10,142,791 B2
`(10) Patent No.:
`Aksamit
`(45) Date of Patent:
`*Nov. 27, 2018
`
`(54) METHOD AND SYSTEM FOR CONTEXT
`AWARENESS OF A MOBILE DEVICE
`
`(56)
`
`References Cited
`U. S. PATENT DOCUMENTS
`
`(71)
`
`Applicant: BINARTECH SP. Z 0.0., Opole (PL)
`
`(72)
`
`Inventor: Pawel Aksamit, Opole (PL)
`
`(73)
`
`Assignee: Binartech Sp. z 0.0., Opole (PL)
`
`(*)
`
`Notice:
`
`Subject to any disclaimer, the term of this
`patent is extended or adjusted under 35
`U.S.C. 154(b) by 0 days.
`
`This patent is subject to a terminal dis-
`claimer.
`
`(21)
`
`Appl. No.: 15/719,881
`
`(22)
`
`Filed:
`
`Sep. 29, 2017
`
`7,203,635 B2*
`
`7,778,632 B2
`
`4/2007 Oliver .................. G06K 9/6293
`345/156
`
`8/2010 Kurlander et al.
`(Continued)
`
`FOREIGN PATENT DOCUMENTS
`
`W0
`WO
`
`2010133770 A1
`2012001215 A1
`
`11/2010
`1/2012
`
`OTHER PUBLICATIONS
`
`Kang, Sungwoo et 31.,“M0biC0n: Mobile Context Monitoring Plat-
`form for Sensor-Rich Dynamic Environments”, Dec. 1, 2010, 14
`pages, retrieved from <http://www.csc.lsu.edu/~iyengar/final-papers/
`CACMim.pdf>.
`
`Prior Publication Data
`
`(Continued)
`
`(65)
`
`(63)
`
`US 2018/0027380 A1
`
`Jan. 25, 2018
`
`Related U.S. Application Data
`
`Continuation of application No. 15/377,414, filed on
`Dec. 13, 2016, now Pat. No. 9,807,564, which is a
`(Continued)
`
`(30)
`
`Foreign Application Priority Data
`
`Feb. 17, 2012
`
`(PL) ...................................... P 398136
`
`(51)
`
`Int. Cl.
`H04W 24/00
`H04W 4/02
`
`(2009.01)
`(2018.01)
`(Continued)
`
`(52) U.S. Cl.
`CPC ........... H04W 4/027 (2013.01); H04W4/025
`(2013.01); H04W24/02 (2013.01);
`(Continued)
`Field of Classification Search
`CPC ..... H04W 4/027; H04W 4/025; H04W 24/02;
`H04W 24/00; H04W 52/0254;
`(Continued)
`
`(58)
`
`101
`
`10 ‘
`
`105 ”102
`
`Primary Examiner 7 Mong-Thuy Tran
`(74) Attorney, Agent, or Firm 7 Lewis & Reese, PLLC
`
`(57)
`
`ABSTRACT
`
`A method for detecting a context of a mobile device (100)
`equipped with sensors (111, 121, 122, 131) and a context
`detection module (109) in which the sensors (111, 121, 122,
`131) are assigned to at least two groups (110, 120, 30), each
`of which comprises at least one sensor (111, 121, 122, 131),
`and each group (110, 120, 130) is allocated a group classifier
`(116, 126, 136) 10 adapted to detect,
`in a form of a
`classification result, currently identified, by means of a given
`classifier, context of the device (100) based on indications of
`the sensors (111, 121, 122, 131) belonging to the given
`group, characterized in that with a use of the context
`detection module, whereas the groups (110, 102, 130) of
`sensors are ordered hierarchically, and the device context is
`detected 1 by reading a classification result indicated by the
`classifier (116, 126, 136) of the currently active group,
`wherein in case of detection of an identified context in the
`
`active group, switching on power supply of the sensors and
`activating classification in a group (110, 120, 130) with a
`level higher by one level and reading the context indicated
`(Continued)
`
`Context
`detecimn
`module
`
`Non—Volatile
`Memory
`
`
`
`
`
`Case 6:20-cv-01063-ADA Document 1-1 Filed 11/17/20 Page 3 of 16
`Case 6:20-cv-01063-ADA Document 1-1 Filed 11/17/20 Page 3 of 16
`
`US 10,142,791 B2
`
`Page 2
`
`by said group’s classifier, wherein based on the 20 results of
`the classification indicated by the higher groups’ classifiers
`(116, 26, 136), executing adaptation of the configuration of
`lower groups’ classifiers (116, 126, 136).
`
`20 Claims, 5 Drawing Sheets
`
`Related US. Application Data
`
`continuation of application No. 14/745,433, filed on
`Jun. 21, 2015, now Pat. No. 9,549,292, which is a
`continuation of application No. 14/346,985, filed as
`application No. PCT/EP2013/052187 on Feb. 5,
`2013, now Pat. No. 9,107,093.
`
`Int. Cl.
`
`(51)
`
`H04W 52/02
`H04 W 24/02
`H04M 1/725
`G06F 3/01
`H04L 29/08
`US. Cl.
`
`(2009.01)
`(2009.01)
`(2006.01)
`(2006.01)
`(2006.01)
`
`CPC ......... H04W52/0254 (2013.01); G06F 3/017
`(2013.01); H04L 67/22 (2013.01); H04M
`[/72569 (2013.01); H04W 24/00 (2013.01);
`2021) 70/00 (2018.01); Y02D 70/26 (2018.01)
`Field of Classification Search
`CPC ...... H05K 999/99; Y02D 70/26; Y02D 70/00;
`G06F 3/017; H04L 67/22; H04M 1/72569
`USPC ....................................................... 455/456.1
`
`(52)
`
`(58)
`
`(56)
`
`References Cited
`U.S. PATENT DOCUMENTS
`
`7,986,914 B1
`8,417,296 B2
`9,107,093 B2
`2002/0128000 A1
`2003/0139654 A1
`2003/0197597 A1
`2004/0002838 A1
`2005/0255874 A1
`2006/0119508 A1
`2007/0100480 A1
`2008/0143518 A1
`2008/0195584 A1
`2008/0235318 A1*
`
`2009/0128286 A1
`2009/0221279 A1
`2010/0048256 A1
`2010/0075652 A1
`2010/0302028 A1
`2010/0306711 A1
`2011/0243448 A1
`2012/0059780 A1
`2012/0100895 A1
`2012/0185419 A1
`2013/0158686 A1
`2013/0173513 A1
`2013/0238535 A1
`
`7/2011 Henry, Jr. et a1.
`4/2013 Caballero et a1.
`8/2015 Aksamit
`9/2002 do Nascimento, Jr.
`7/2003 Kim et a1.
`10/2003 Bahl et a1.
`1/2004 Oliver et a1.
`11/2005 Stewart-Baxter et a1.
`6/2006 Miller
`5/2007 Sinclair et a1.
`6/2008 Aaron
`8/2008 Nath et a1.
`9/2008 Khosla ................. G06K 9/6292
`709/201
`
`5/2009 Vitito
`9/2009 Rutledge
`2/2010 Kluppi et a1.
`3/2010 Keskar et a1.
`12/2010 Desai et a1.
`12/2010 Kahn et a1.
`10/2011 Kawabuchi
`3/2012 Kononen et a1.
`4/2012 Priyantha et a1.
`7/2012 Kuhn et a1.
`6/2013 Zhang et a1.
`7/2013 Chu et a1.
`9/2013 Leppanen et a1.
`
`OTHER PUBLICATIONS
`
`Wang, Yi et a1., “A Framework of Energy Efiicient Mobile Sensing
`for Automatic User State Recognition,” Proceesing of 7th Annual
`International Conference on Mobile Systems Applications and
`Services (MobiSys), 2009, pp. 179-192.
`
`See application file for complete search history.
`
`* cited by examiner
`
`
`
`Case 6:20-cv-01063-ADA Document 1-1 Filed 11/17/20 Page 4 of 16
`Case 6:20-cv-01063-ADA Document 1-1 Filed 11/17/20 Page 4 of 16
`
`U.S. Patent
`
`0N
`
`001
`
`US 10,142,791 B2
`
`t.h.w"mgmhnS.
`5nfn0.1n333
`mmanna
`anragga
`v.wxmwcco
`
`
`we“
`
`.525
`ii...
`
`
`
`Case 6:20-cv-01063-ADA Document 1-1 Filed 11/17/20 Page 5 of 16
`Case 6:20-cv-01063-ADA Document 1-1 Filed 11/17/20 Page 5 of 16
`
`U.S. Patent
`
`Nov. 27,2018
`
`Sheet 2 of5
`
`US 10,142,791 132
`
`111111.131
`Power—
`Fewer—
`on
`,
`witch .....
`
`Group 2
`
`‘
`
`Graup 1
`
`.....
`
`
`
`
`1.15.3 ...................................
`Features
`; Geteiminatien
`g
`Moduie ”i
`
`5
`
`5
`
`Features
`Qeterminaiion
`Mamie 2
`
`...............................................
`E
`Features
`Determinaiéen
`5 Moduie 3
`
`
`
`Feaéuree
`vector
`3
`”536
`i m9 ...............................
`'
`____ Ciassiiier
`3
`
`3
`3
`E
`E
`
`a
`3 3"
`a 3:35;
`a g
`g
`
`VVVV
`
`
`Features;
`Features
`water
`water
`3
`2
`12
`1
`3
`1’3
`W ,,...........................................................
`_
`________
`Ciassifier
`3
`
`a
`i
`1
`
`Ciaseificafien
`g
`E
`rasuifl
`
`5
`~
`.1
`__
`.
`a
`g
`Ciasséfication
`g
`resuét?
`'
`
`
`:
`5
`g
`
`
`
`
`_______
`
`109 g
`x\
`5
`
`Comexi detection moduie
`
`
`
`Case 6:20-cv-01063-ADA Document 1-1 Filed 11/17/20 Page 6 of 16
`Case 6:20-cv-01063-ADA Document 1-1 Filed 11/17/20 Page 6 of 16
`
`U.S. Patent
`
`Nov. 27, 2018
`
`Sheet 3 of 5
`
`US 10,142,791 B2
`
`
`EemmommmmmmuEmwmomwwmmfi6%$ng"6ngmmE19$”Emmi”aanewmgwamwgumm
`
`
`max/mumEmcowumomammamag?EmWsowhowcmmmgWe52,3Ewe/m.113...;ca@038mg3E258,£31wa
`
`
`
`_gem5%EggmgEemmofimmmacommammmma33%new
`
`
`@3805EngowgmmEmB53%Egg
`new.W
`
`
` m2,9mcmmags,meEvamam5%
`mm»02
`
`cawmgmmmmmméwmmogwcm
`
`
`
`
`
`fiemwwmmmmmmmo$5me$5,61mem...............................VQN..................................WEmmmuommcamam
`
`02
`
`mm,”..
`
`magma3E58”Em
`
`mam
`
`
`
`
`
`
`
`3new$5”Ego“6mg35$5”$.52EmummmmwwmgBEommfimmmwgmmmwgu.5ngmamum
`
`
`.,ma39m$95”EggE.aa
`
`3&9$ng.6mxfiwmmfiwecammammmDEEare“;E33%@1mem35.935$.mg88$3$41
`
`mam
`
`m,ma
`
`
`
`
`
`
`
`
`
`
`Case 6:20-cv-01063-ADA Document 1-1 Filed 11/17/20 Page 7 of 16
`Case 6:20-cv-01063-ADA Document 1-1 Filed 11/17/20 Page 7 of 16
`
`U.S. Patent
`
`Nov. 27,2018
`
`Sheet 4 of5
`
`US 10,142,791 132
`
`145
`
`N
`
`F 1
`’1
`r
`32:53
`"
`
`
`
`Senscrs
`
`centre;
`
`Classification
`
`reeuii
`
`‘
`
`
`
`
`A set 0?
`A set of
`pesétive paiiema
`negative patterns
`
`
`
`
`
`3
`
`-
`
`
`
`Adaptatim moduée
`
`Ciassifécation Resufis
`
`
`
`
`
`01‘ higher ievei group(s}
`
`Fig, 4
`
`
`
`Case 6:20-cv-01063-ADA Document 1-1 Filed 11/17/20 Page 8 of 16
`Case 6:20-cv-01063-ADA Document 1-1 Filed 11/17/20 Page 8 of 16
`
`U.S. Patent
`
`Nov. 27,2018
`
`Sheet 5 of5
`
`US 10,142,791 132
`
`Load a features vector
`
`subject in ciassificatiai‘i
`andaciassificatimresuii
`
`‘
`
`(ResuitA)
`
`A pesiiive, 8 negative
`
`Load ciaSSifisa‘ion resuit
`
`
`
`of a ciassifier @f a higher
`ievei gmup (Remit: B)
`
`
`
`
`3
`of positive patterns
`
`A negative; 8 gosiiive
`
`3135
`
`um.
`
`304
`
`i
`
`Add ihe feaiures vecmr
`
`
`
`‘10 a sefi of positive
`pafiems
`
`
`_
`
`5
`
`Remeve the features vecmf I
`
`imm a set of negative
`gatiarns and add ta 8 set
`
`A pasitive B negative
`
`Remove the features vectar
`.
`.
`{ram 23 set 91‘ gmaitive
`patterns and add is a $91
`
`(if negative gaitema
`
`A negative, 8 negative
`
`
`
`to a set of negative
`eatiems
`
`
`Add the features veciei
`
`Fig. 5
`
`
`
`Case 6:20-cv-01063-ADA Document 1-1 Filed 11/17/20 Page 9 of 16
`Case 6:20-cv-01063-ADA Document 1-1 Filed 11/17/20 Page 9 of 16
`
`US 10,142,791 B2
`
`1
`METHOD AND SYSTEM FOR CONTEXT
`AWARENESS OF A MOBILE DEVICE
`
`CROSS-REFERENCE TO RELATED
`APPLICATIONS
`
`This is a continuation of US. patent application Ser. No.
`15/377,414, filed Dec. 13, 2016 (now allowed), which is a
`continuation of US. patent application Ser. No. 14/745,433,
`filed Jun. 21, 2015 (now US. Pat. No. 9,549,292), which is
`a continuation of US. patent application Ser. No. 14/346,
`985, filed Mar. 25, 2014 (now US. Pat. No. 9,107,093),
`which is a national stage entry of PCT Patent Application
`Serial No. PCT/EP2013/052187, filed Feb. 5, 2013, which
`claims priority to Polish Patent Application No. P398136,
`filed Feb. 17, 2012. Priority is claimed to these applications,
`and these applications are incorporated herein by reference
`in their entireties.
`
`FIELD OF THE INVENTION
`
`The present invention relates to a method for detecting
`context of a mobile device and to a mobile device having a
`context detection module, especially to detect
`that
`the
`mobile device is located in a moving vehicle.
`
`BACKGROUND
`
`A desirable feature of mobile devices, such as mobile
`phones,
`laptops, PDAs,
`tablets, watches, music players,
`satellite navigation devices, cameras,
`is awareness of the
`device regarding the environment in which it is located, the
`current activity of the user and/or the circumstances in
`which the user finds himself, which together are called the
`“context” in the reminder of this description. Awareness of
`the context can help to improve the usability of the device,
`as well as the comfort and safety of its use.
`Today’s mobile devices are equipped with a large number
`of different types of sensors, which allow, inter alia, auto-
`matic activation and deactivation of individual functions or
`
`to change configuration of mobile devices, depending on the
`context. In most cases, the sensors indicate the context only
`indirectly. Therefore, there are carried out in the industry
`intense developments of effective and efficient methods for
`determining the context based on signals from various types
`of sensors.
`
`A special case of context is a situation when a user of a
`mobile device drives a vehicle, such as a car. It is inadvisable
`in such circumstances, and in many countries prohibited by
`law to use such devices’ functions so as make and receive
`
`voice calls and send and receive text messages. On the other
`hand, in those circumstances, it might be advantageous to
`activate other functions, such as navigation or download
`from an external database of information on known hazards
`
`on the roads and to inform the user of approaching them,
`with a message of a tone, voice, visual, or any combination
`thereof.
`
`One of the known ways of detecting that the mobile
`device is located in a moving vehicle is to determine its
`approximate position on the basis of signals of base stations
`of cellular telephony and calculating average speed of the
`device based on the change of thus determined position in
`time. Position the device specified according to the afore-
`mentioned method is typically uncertainty in the order of
`few hundred meters to several kilometers. The greater the
`movement of the device in time, the greater the certainty of
`the weighted average speed. In a practical use this means a
`
`10
`
`15
`
`20
`
`25
`
`30
`
`35
`
`40
`
`45
`
`50
`
`55
`
`60
`
`65
`
`2
`
`compromise between a delay in detection of movement in a
`vehicle of up to several minutes in the teens, and acceptance
`of frequent false classifications indicating the movement of
`the device in a vehicle at times when this is not actually
`happening.
`Another well-known and used method of detecting that a
`mobile device is in a moving vehicle is to analyze the
`movement speed of the device using a global satellite
`positioning system, which provides much more accurate
`positioning than the signals of mobile communications base
`stations. Receivers of this type, however, consume signifi-
`cant amounts of energy, which significantly reduces the
`operating time of the mobile device using battery, often to
`such levels that are unacceptable to the user. Power con-
`sumption is reduced in many cases by less frequent switch-
`ing the receiver on and less frequent positioning. However,
`this leads to a delay in detection of the identified context.
`Given the relatively small capacity of the batteries used in
`mobile devices, a compromise that can be achieved in this
`method between the movement detection delay and energy
`consumption can be unsatisfactory for many applications.
`The patent application US. 2002/0128000 A1 discloses a
`system for use in a mobile communication device. The
`system includes a subsystem used for detection of move-
`ment of a device in a vehicle. Detection is performed by
`measuring the average speed of movement of the device,
`determined on the basis of signals from mobile telephony
`base stations or with a use of global satellite positioning
`system receiver.
`An alternative and used on a industrial scale method for
`
`detecting use of a mobile device in a vehicle is to use
`vehicle-mounted short-range radio transmitter. The trans-
`mitter can be optionally activated only during operation of
`the vehicle, such as when the vehicle engine is running. The
`mobile device is equipped with a receiver compatible with
`the transmitter. It is assumed that the mobile device is used
`
`in a vehicle, when it is in range of the vehicle mounted
`transmitter. The disadvantage of this method is the necessity
`of an additional transmitter in the vehicle, while in the case
`of universal transceiver module, which is an equipment of
`the most modern mobile phones and some modern carsiat
`least a mindful configuration by a user. The configuration in
`this case is based on searching for a signal of the built-in
`transmitter of the vehicle, and storing a network address in
`the mobile device for future, unambiguous identification.
`Patent application US. 2005/0255874 A1 discloses a
`system and method for detecting movement in a vehicle,
`which consists of a vehicle-mounted radio transmitter hav-
`
`ing low range, activated at a time when the vehicle is in use,
`and a mobile device equipped with a radio receiver. The
`mobile device detects that it is in the vehicle being used
`based on proximity of the signal of the transmitter mounted
`in the vehicle.
`
`There are also ways of specifying a device context based
`one analysis of signals from different sensors, such as
`microphones, accelerometers, light sensors, magnetic field
`sensors, compasses, cameras and other. The information
`contained in the signals from the individual sensors are
`usually insufficient in order to determine, with a satisfacto-
`rily high degree of certainty, the device’s context. Therefore,
`most of the known methods consist of parallel analysis of
`signals from multiple sensors. This leads to a substantial
`increase in the effectiveness of the classification, which,
`however,
`is achieved at the expense of increased power
`consumption, and thus a shorter time of operation of a
`battery-powered mobile device. The problem of reducing
`the energy consumption, if at all addressed in descriptions of
`
`
`
`Case 6:20-cv-01063-ADA Document 1-1 Filed 11/17/20 Page 10 of 16
`Case 6:20-cv-01063-ADA Document 1-1 Filed 11/17/20 Page 10 of 16
`
`US 10,142,791 B2
`
`3
`is typically solved by less frequent
`individual methods,
`switching the sensors on and less frequent sampling. This
`leads, however, to extension of time after which a change of
`device context is detected.
`The effectiveness of the classification of the signals from
`the individual sensors or groups of sensors, is also increased
`by the mechanisms of adaptive customization of classifiers’
`configuration based on feedback provided by users. They
`require the user to go through a conscious device learning
`process. There are also known methods for adaptation of
`classifiers determining the context based on the analysis of
`natural user interaction with the device, from which a real
`context at the time is determined. In the case of detecting
`movement in a vehicle, it may be a connection of the mobile
`phone to a speakerphone or executing applications for
`navigation. In this case, the adaptation of the classifiers’
`configuration, however, is much slower than in the case of
`intentional learning by the user. In the meantime, before the
`classifier’s configuration adapts to the specific features indi-
`cating the contexts in which the user uses the device, the
`classification results can be far from satisfactory.
`Patent application WO 2010/133770 A1 discloses a
`method of detecting a mobile device context. The method is
`based on retrieval of data from sensors, the data indicating
`a context in which the device and its user are found in,
`determining from the data their features, subjecting these
`features to a classification using an adaptive linear classifier
`and an adaptation of the classifier’s configuration based on
`the designated features and feedback provided by the user of
`the device. Low complexity of the proposed classifier affects
`the low energy consumption compared to more complex
`methods of classification. Adaptation of configuration of the
`classifier based on information derived from user feedback,
`in turn increases the efficiency of classification. Feedback
`used to amend classifier’ s settings is provided directly by the
`user,
`indicating the correct classification result or voting
`positively or negatively on the result returned by the clas-
`sifier, or indirectly inferred from actions taken by the user or
`the absence thereof.
`
`Patent application publication USZOO9128286 discloses a
`system for controlling the use of electronic devices within an
`automobile includes a control module integrated into the
`operation of the automobile, wherein the control module is
`linked to the electrical system of the automobile and is
`controlled, monitored and updated via a graphical user
`interface of the automobile. The control module includes a
`
`mechanism for identifying use of electronic devices within
`the automobile, determining whether the usage is permitted
`and preventing usage of the electronic device if it is deter-
`mined the usage is not permitted.
`Patent application publication U82002128000 discloses a
`system for use with a mobile communication unit includes
`a service device configured to determine at least one service
`to be applied to affect at least one of incoming and outgoing
`communications to and from, respectively, the mobile com-
`munication unit,
`the service device being configured to
`determine the at least one service to be applied dependent
`upon a location of the mobile communication unit.
`It would be desirable to define a method of detecting
`context of a mobile device and a mobile device having a
`module that enables detection of its context, which will
`reduce at least some of the disadvantages present in the prior
`art solution, and which will provide for a use of an alterna-
`tive mechanism for the detection of context.
`
`DISCLOSURE
`
`The invention relates to a method for detecting a context
`of a mobile device equipped with sensors and a context
`
`10
`
`15
`
`20
`
`25
`
`30
`
`35
`
`40
`
`45
`
`50
`
`55
`
`60
`
`65
`
`4
`
`detection module in which the sensors are assigned to at
`least two groups, each of which comprises at
`least one
`sensor, and each group is allocated a group classifier adapted
`to detect,
`in a form of a classification result, currently
`identified, by means of a given classifier, context of the
`device based on the indications of the sensors belonging to
`the given group. With a use of a context detection module,
`the groups of sensors are ordered hierarchically, the device
`context is detected by reading a classification result indi-
`cated by the classifier of the currently active group, wherein
`in case of detection of an identified context in the active
`
`group, there is switched on power supply of the sensors and
`there is activated classification in a group with a higher level
`and there is read the context
`indicated by said group’s
`classifier, wherein based on the results of the classification
`indicated by the higher groups classifiers there is made an
`adaptation of the configuration of lower groups’ classifiers.
`Preferably, if the result of the classification indicated by
`a top level group classifier is a positive result of the
`classification, there is executed at least one operation from
`the set, which comprises: adapting user interface of the
`mobile device, changing of state or mode of an operating
`application of the mobile device, execution of an applica-
`tion, closing of an application, activation of a function of the
`mobile device, mobile device’s
`function deactivation,
`changing of the configuration of the mobile device, playback
`of a sound signal, establishing a voice call, sending of a text
`message, sending of a graphical message, establishing a data
`connection, data transmission.
`Preferably, at least one group of sensors comprises a
`sensor from a set, which includes: an accelerometer, a
`microphone, a magnetic field sensor, a compass, a light
`sensor, a camera, a signal strength sensor of mobile tele-
`phony base stations, a proximity sensor of a radio transmitter
`or relay, a receiver of satellite geographic positioning sys-
`tem.
`
`the satellite geographic positioning system
`Preferably,
`receiver belongs to the group the highest level.
`Preferably, the sensor groups are ordered hierarchically
`such that the total amount of energy required to determine a
`classification result in lower levels groups is less than the
`amount of energy required to determine the result of the
`classification in groups of higher levels.
`Preferably, the sensors groups are ordered hierarchically
`in such a way that the result of the classification in groups
`of higher levels has lower uncertainty than the classification
`result of the lower levels groups.
`Preferably, switching the power on for at least one sensor
`is based on a change of its mode of operation from a more
`energy efficient to a less energy efficient.
`Preferably, substantially in parallel with powering on a
`higher level group of sensors there are powered off sensors
`of a lower level group.
`Preferably, switching the power off for at least one sensor
`is based on a change of its mode of operation from a less
`energy efficient to a more energy efficient.
`Preferably, at least one sensor is switched to a reduced
`power consumption after reading indication of a given
`sensor.
`
`Preferably, at least one of the sensors belonging to the
`lowest
`level group has power supply switched on in a
`constant mode.
`
`Preferably, there is switched on at least one of the sensors
`belonging to the lowest level group, periodically or accord-
`ing to a fixed schedule.
`Preferably, the continuous powering or period or power
`on schedule of at least one sensor depend on the internal
`
`
`
`Case 6:20-cv-01063-ADA Document 1-1 Filed 11/17/20 Page 11 of 16
`Case 6:20-cv-01063-ADA Document 1-1 Filed 11/17/20 Page 11 of 16
`
`US 10,142,791 B2
`
`5
`state of the mobile device, on the configuration of the device
`selected by the user, on an entry in a user’s calendar or on
`the currently specified device’s context.
`Preferably, indications of the sensors assigned to a given
`group, are read by means of a module for determining
`features assigned to the given group, and adapted for deter-
`mining a features vector on the basis of readings of the
`Sensors belonging to the given group.
`Preferably, the features vector further includes readings of
`a sensor belonging to a group of lower level.
`Preferably, in at least one classifier there is stored a set of
`positive patterns comprising features vectors indicating a
`positive result of the classification, and a set of negative
`patterns comprising features vectors indicating a negative
`result of the classification.
`
`Preferably, at least one classifier is adapted to determine
`the context of the device based on the formula of “k nearest
`
`neighbours”.
`The invention also provides a mobile device equipped
`with sensors and a context detection module, in which the
`sensors are assigned to at least two groups, each of which
`comprises at least one sensor, whereas the context detection
`module comprises
`classifiers associated with specific
`groups, wherein each classifier is adapted to detect, in a form
`of a result of the classification, the currently identified, by a
`given context classifier, context of the device based on
`readings from the sensors belonging to the given group. The
`groups of sensors are arranged hierarchically, and the con-
`text detection module is adapted to checking the context of
`the device by reading the result of the classification indi-
`cated by the currently active group’s classifier, whereas each
`classifier comprises a module for determining the classifi-
`cation coupled to a power supply controller, which is acti-
`vated by a signal indicating a detection by the module for
`determining the classification of the identified context in the
`given group and adapted to power on a sensors group having
`a higher level, in order to read the context indicated by the
`group’s classifier, whereas the classifiers further comprise an
`adaptation module configured to adapt the configuration of
`the classifier based on the results of classifications indicated
`
`by the classifiers of higher level groups.
`Preferably, the device is adapted to execute, if the result
`of the classification indicated by a top level group classifier
`is a positive result of the classification, at least one operation
`from the set, which comprises: user interface adaptation of
`the mobile device, a change of state or mode of an operating
`application of the mobile device, execution of an applica-
`tion, closing of an application, activation of a function of the
`mobile device, mobile device’s
`function deactivation,
`changing of the configuration of the mobile device, playback
`of a sound signal, establishing a voice call, sending of a text
`message, sending of a graphical message, establishing a data
`connection, data transmission.
`Preferably, at least one group of sensors comprises a
`sensor from a set, which includes: an accelerometer, a
`microphone, a magnetic field sensor, a compass, a light
`sensor, a camera, a signal strength sensor of mobile tele-
`phony base stations, a proximity sensor of a radio transmitter
`or relay, a receiver of satellite geographic positioning sys-
`tem.
`
`the satellite geographic positioning system
`Preferably,
`receiver belongs to the group the highest level.
`Preferably, the sensor groups are ordered hierarchically
`such that the total amount of energy required to determine a
`classification result in lower levels groups is less than the
`amount of energy required to determine the result of the
`classification in groups of higher levels.
`
`6
`Preferably, the sensors groups are ordered hierarchically
`in such a way that the result of the classification in groups
`of higher levels has lower uncertainty than the classification
`result of the lower levels groups.
`Preferably,
`the power supply controller is adapted to
`switching the power on for at least one sensor by changing
`its mode of operation from a more energy efficient to a less
`energy efficient.
`Preferably,
`the power supply controller is adapted to,
`substantially in parallel with powering on a higher level
`group of sensors, power off sensors of a lower level group.
`Preferably, switching the power off for at least one sensor
`is based on a change of its mode of operation from a less
`energy efficient to a more energy efficient.
`Preferably, at least one sensor is switched to a reduced
`power consumption after reading indication of a given
`sensor.
`
`10
`
`15
`
`Preferably, at least one of the sensors belonging to the
`lowest
`level group has power supply switched on in a
`constant mode.
`
`20
`
`Preferably, there is switched on at least one of the sensors
`belonging to the lowest level group, periodically or accord-
`ing to a fixed schedule.
`Preferably, the continuous powering or period or power
`on schedule of at least one sensor depend on the internal
`state of the mobile device, on the configuration of the device
`selected by the user, on an entry in a user’s calendar or on
`the currently specified device’s context.
`Preferably, the context detection module is equipped with
`modules for determining features, of which each is con-
`nected to a given group and is configured to determine a
`features vector on the basis of readings of the sensors
`belonging to the given group.
`Preferably, the features vector further includes readings of
`a sensor belonging to a group of lower level.
`Preferably, in at least one classifier there is stored a set of
`positive patterns comprising features vectors indicating a
`positive result of the classification, and a set of negative
`patterns comprising features vectors indicating a negative
`result of the classification.
`
`Preferably, at least one classifier is adapted to determine
`the context of the device based on the formula of “k nearest
`
`neighbours”.
`An advantage of the invention is that it enables effective
`detection of context by powering on a minimum number of
`sensors, as rarely as possible and for the shortest period of
`time. A smaller number of powered sensors and shorter time
`of reading signal obtained from them means, however, a
`smaller amount of information and the associated higher
`uncertainty of the obtained classification result. It is there-
`fore advantageous to take into account as much information
`from as many independent sensors, advantageously retum-
`ing signal related to different physical phenomena (e.g. a
`microphone-acoustic wave and an accelerometer-accelera-
`tion). Thus, a solution consisting of grouping sensors and of
`powering on subsequent classification levels only after a
`positive result returned by the preceding levels, is an advan-
`tageous compromise between the first and the second option.
`In addition, it is to be noted that due to the huge variety of
`circumstances in which a used may be, it is very difficult to
`subject the individual classifiers to a single-time learning
`process, which would result
`in classifying correctly the
`signals read from the sensors in all circumstances. However,
`if the mobile device’s user is found in circumstances that the
`
`classifier will regularly classify incorrectly, it can lead to
`behaviour
`inconsistent with expectations regarding the
`device or regular powering on sensors belonging to the
`
`25
`
`30
`
`35
`
`40
`
`45
`
`50
`
`55
`
`60
`
`65
`
`
`
`Case 6:20-cv-01063-ADA Document 1-1 Filed 11/17/20 Page 12 of 16
`Case 6:20-cv-01063-ADA Document 1-1 Filed 11/17/20 Page 12 of 16
`
`US 10,142,791 B2
`
`7
`groups of high levels and subjecting to the classification the
`signal read from them, which in turn will lead to faster than
`desired use of battery energy. Consequently, the key is an
`adaptation mechanism for lower level classifier configura-
`tion based on classification results returned by classifiers of
`a higher level. Significantly, it is a mechanism which does
`not require any action from the user. Because of their ability
`to adapt, regardless of their initial configuration, the lower
`level classifiers adapt their configuration in such a way as to
`return results the most convergent with the results returned
`by the classifiers of higher levels in the same circumstances.
`After a certain number of adaptation cycles, dependent on
`the given circumstances and the initial configuration of the
`classifier, there ceases to be necessary enabling the sensors
`belonging to higher level groups, and subjecting the signals
`originating from them to classification in order to obtain a
`correct result for the given circumstances.
`
`BRIEF DESCRIPTION OF THE DRAWINGS
`
`The object of the solution h