Abstract: Systems, methods, and program products for providing services to a user by a mobile device based on the user's daily routine of movement. The mobile device determines whether a location cluster indicates a significant location for the user based on one or more hints that indicate an interest of the user in locations in the cluster. The mobile device can perform adaptive clustering to determine a size of area of the significant location based on how multiple locations converge in the location cluster. The mobile device can provide location-based services for calendar items, including predicting a time of arrival at an estimated location of a calendar item. The mobile device can provide various services related to a location of the mobile device or a significant location of the user through an application programming interface (API).

Abstract: Crowds of people within an environment can be modeled in real time. A multitude of mobile devices located within an environment can periodically transmit their geographical locations over networks to a remote server. The remote server can use these geographical locations to generate a current real-time model of a crowd of people who possess the mobile devices that transmitted the geographical locations. The remote server can transmit the model over networks back to the mobile devices. The mobile devices can use the received model to present useful information to the users of those mobile devices.

Abstract: A proximity fence can be a location-agnostic fence defined by signal sources having no geographic location information. The proximity fence can correspond to a group of signal sources instead of a point location fixed to latitude and longitude coordinates. A signal source can be a radio frequency (RF) transmitter broadcasting a beacon signal. The beacon signal can include a payload that includes an identifier indicating a category to which the signal source belongs, and one or more labels indicating one or more subcategories to which the signal source belongs. The proximity fence defined by the group of signal sources can trigger different functions of application programs associated with the proximity fence on a mobile device, when the mobile device moves within the proximity fence and enters and exits different parts of the proximity fence corresponding to the different subcategories.

Abstract: An event can be detected by an input device. The event may be determined to be a triggering event by comparing the event to a group of triggering events. A first prediction model corresponding to the event is then selected. Contextual information about the device specifying one or more properties of the computing device in a first context is then received, and a set of one or more applications is identified. The set of one or more applications may have at least a threshold probability of being accessed by the user when the event occurs in the first context. Thereafter, a user interface is provided to a user for interacting with the set of one or more applications.

Abstract: Methods, program products, and systems for reducing a location search space are described. A mobile device, when arriving at a venue, can determine a location of the mobile device using signals from one or more signal sources associated with the venue. The mobile device can use a coarse location estimator to estimate a coarse location of the mobile device at the venue. The mobile device can request, from a server, detailed location data associated with the coarse location. The detailed location data can include location fingerprint data associated with a portion of the venue that includes the coarse location. The mobile device can determine an estimated location that has finer granularity than the coarse location using the location fingerprint data.

Abstract: Methods and computer products can provide personalized content based on historical interaction with a mobile device. A computing device can receive information about a user interaction with an application running on the mobile device at a first time and location. A type of the application can be identified by parsing a description of the application (e.g., using a natural language processing algorithm). An affinity model can be generated that associates the type of the application with the first time and/or location. At a second time and location, it can be determined that the second time corresponds to the first time and/or that the second location corresponds to the first location. Using the affinity model, the second time and/or location can be associated with the type of the application, and the mobile device may then display content related to the type of the application.

Abstract: Methods, program products, and systems of location estimation using a probability density function are disclosed. In general, in one aspect, a server can estimate an effective altitude of a wireless access gateway using harvested data. The server can harvest location data from multiple mobile devices. The harvested data can include a location of each mobile device and an identifier of a wireless access gateway that is located within a communication range of the mobile device. The server can calculate an effective altitude of the wireless access gateway using a probability density function of the harvested data. The probability density function can be a sufficient statistic of the received set of location coordinates for calculating an effective altitude of the wireless access gateway. The server can send the effective altitude of the wireless access gateway to other mobile devices for estimating altitudes of the other mobile devices.

Abstract: Surveying techniques for generating location fingerprint data are described. A mobile device can survey a venue by measuring, at multiple locations at the venue, signals from one or more signal sources. At each location, the mobile device can take multiple measurements of signals. The mobile device can take each measurement at a distinct orientation. The measurements can be used to determine expected measurements of the signals at the venue. Differences between the multiple measurements of signals can be used to determine a variance of the expected measurements. The expected measurements and variance can be designated as location fingerprint data for the venue. The location fingerprint data can be used by mobile devices for determining a location at the venue.

Abstract: An event can be detected by an input device. The event may be determined to be a triggering event by comparing the event to a group of triggering events. A first prediction model corresponding to the event is then selected. Contextual information about the device specifying one or more properties of the computing device in a first context is then received, and a set of one or more applications is identified. The set of one or more applications may have at least a threshold probability of being accessed by the user when the event occurs in the first context. Thereafter, a user interface is provided to a user for interacting with the set of one or more applications.

Abstract: Techniques of category-based fence are described. A category-based fence can correspond to a group of signal sources instead of a point location fixed to latitude and longitude coordinates. The group of signal sources can represent a category of entities, e.g., a particular business chain. The signal sources can be distributed to multiple discrete locations. A category-based fence associated with the group, accordingly, can correspond to multiple locations instead of a single point location and a radius. Each signal source in the group can be associated with a category identifier unique to the group and uniform among signal sources in the group. The category identifier can be programmed into each signal source. A mobile device can enter the category-based fence by entering any of the discrete locations when the mobile device detects the signal identifier. The mobile device can then execute an application program associated with the category-based fence.

Abstract: Systems, methods and computer-readable mediums are disclosed for rapid reconfiguration of device location systems. In some implementations, a method comprises: determining, by a server computer, a fault in global location data, the server computer configured to periodically communicate with client devices over a network, the global location data being applicable to multiple client devices and independent of any particular location of any particular client device; determining, by the server computer, a potential impact of the faulty global location data on location systems implemented on the client devices; generating, by the server computer, a reconfiguration file including reconfiguration data for reconfiguring the location systems on the client devices to adapt to the faulty global location data; and automatically sending, by the server computer, the reconfiguration file to the client devices.

Abstract: An event can be detected by an input device. The event may be determined to be a triggering event by comparing the event to a group of triggering events. A first prediction model corresponding to the event is then selected. Contextual information about the device specifying one or more properties of the computing device in a first context is then received, and a set of one or more applications is identified. The set of one or more applications may have at least a threshold probability of being accessed by the user when the event occurs in the first context. Thereafter, a user interface is provided to a user for interacting with the set of one or more applications.

Abstract: A mobile device can provide predictive user assistance based on various sensor readings, independently of or in addition to a location of the mobile device. The mobile device can determine a context of an event. The mobile device can store the context and a label of the event on a storage device. The label can be provided automatically by the mobile device or by the external system without user input. At a later time, the mobile device can match new sensor readings with the stored context. If a match is found, the mobile device can predict that the user is about to perform the action or recognize that the user has performed the action again. The mobile device can perform various operations, including, for example, providing user assistance, based on the prediction or recognition.

Abstract: Systems, methods, and apparatuses are provided for suggesting one or more applications to a user based on an event. A prediction model can correspond to a particular event. The suggested application can be determined using one or more properties of the computing device. For example, a particular sub-model can be generated from a subset of historical data that are about user interactions after occurrences of the event and that are gathered when the device has the one or more properties. A tree of sub-models may be determined corresponding to different contexts of properties of the computing device. And, various criteria can be used to determine when to generate a sub-model, e.g., a confidence level in the sub-model providing a correct prediction in the subset of historical data and an information gain (entropy decrease) in the distribution of the historical data relative to a parent model.

Abstract: Methods, systems and computer program products for user settlement detection are disclosed. A mobile device configured to perform an action immediately upon entering or exiting a geofenced environment can delay performing the action until a user of the mobile device has settled into the environment. The mobile device can detect a settled user state by observing the environment of the mobile device, including measuring one or more environment variables using one or more sensors of the mobile device. The mobile device can detect a settled user state even when the mobile device is in motion. The mobile device can perform the action upon detecting a settled user state.

Abstract: Methods, program products, and systems of using a mobile WAP for location and context purposes are disclosed. In general, in one aspect, a server can estimate an effective location of a wireless access gateway using harvested data. The server can harvest location data from multiple mobile devices. The harvested data can include a location of each mobile device and an identifier of a wireless access gateway that is located within a communication range of the mobile device. In some implementations, the server can identify a mobile wireless access gateway based on a distance comparison. Data indicating the mobility of a wireless access gateway can be used by a mobile device to initiate one or more actions, including managing power of the mobile device, modifying entrance and exit conditions of virtual fences and determining a context of the mobile device.

Abstract: Methods, program products, and systems of location estimation using a probability density function are disclosed. In general, in one aspect, a server can estimate an effective altitude of a wireless access gateway using harvested data. The server can harvest location data from multiple mobile devices. The harvested data can include a location of each mobile device and an identifier of a wireless access gateway that is located within a communication range of the mobile device. The server can calculate an effective altitude of the wireless access gateway using a probability density function of the harvested data. The probability density function can be a sufficient statistic of the received set of location coordinates for calculating an effective altitude of the wireless access gateway. The server can send the effective altitude of the wireless access gateway to other mobile devices for estimating altitudes of the other mobile devices.

Abstract: Implementations are disclosed for obtaining a range state of a device operating in an indoor environment with radio frequency (RF) signal sources. In some implementations, windowed signal measurements obtained from RF signals transmitted by an RF signal source are classified into range classes that are defined by threshold values obtained from a RF signal propagation model. A range class observation is obtained by selecting a range class among a plurality of range classes based on a percentage of a total number of windowed signal measurements that are associated with the range class. The range class observation is provided as input to a state estimator that estimates a range class that accounts for process and/or measurement noise. The output of the state estimator is provided as input to a state machine.

Abstract: Crowds of people within an environment can be modeled in real time. A multitude of mobile devices located within an environment can periodically transmit their geographical locations over networks to a remote server. The remote server can use these geographical locations to generate a current real-time model of a crowd of people who possess the mobile devices that transmitted the geographical locations. The remote server can transmit the model over networks back to the mobile devices. The mobile devices can use the received model to present useful information to the users of those mobile devices.

Abstract: A proximity fence can be a location-agnostic fence defined by signal sources having no geographic location information. The proximity fence can correspond to a group of signal sources instead of a point location fixed to latitude and longitude coordinates. A signal source can be a radio frequency (RF) transmitter broadcasting a beacon signal. The beacon signal can include a payload that includes an identifier indicating a category to which the signal source belongs, and one or more labels indicating one or more subcategories to which the signal source belongs. The proximity fence defined by the group of signal sources can trigger different functions of application programs associated with the proximity fence on a mobile device, when the mobile device moves within the proximity fence and enters and exits different parts of the proximity fence corresponding to the different subcategories.