Context-aware smart alarm application

Context-aware smart alarm application

for smart living

Harshith Allamsetti

School of Computing, Informatics and

Decision Systems Engineering

Arizona State University

Tempe, Arizona – 85281

Email: hallamse@asu.edu

Venkat Sanjay Thotapalli

School of Computing, Informatics and

Decision Systems Engineering

Arizona State University

Tempe, Arizona – 85281

Email: vthotapa@asu.edu

Raghavendra Kolloju

School of Computing, Informatics and

Decision Systems Engineering

Arizona State University

Tempe, Arizona – 85281

Email: rkolloju@asu.edu

Akshit Reddy Gudoor

School of Computing, Informatics and

Decision Systems Engineering

Arizona State University

Tempe, Arizona – 85281

Email: agudoor@asu.edu

Abstract—Traditionally we have used alarm clocks to set time

to help us wake up in the morning.However, some people are

still finding it difficult to wake up in the morning even with

the help of alarm clock as a result of which they are missing

their classes, meetings and even exams. The traditional snooze

option of the alarm clock provides the user more time to sleep

by snoozing the alarm sound for a fixed interval of time. We

plan to develop an intelligent alarm clock system for smart-

phones that delivers alarm sounds by taking into consideration

the user’s sleep patterns, context information such as weather

and social information and events on their calendar for the

following day. Our application aims to provide a smart life by

sensing data collected by multiple sensors on the smart-phone

and collaborating with other applications on the smart phone

to recommend a preferable time for the alarm sound.

1. Introduction

Traditional alarm clock’s have been in use for many

years, yet people are still finding it difficult to wake up in

the morning. Our aim is to provide a one step solution in the

form of an mobile application for people who find it difficult

to wake up to the alarm clock as a result of which they are

missing their meetings and classes. Mobile application de-

velopment in recent times has grown exponentially making

the mobile devices far more capable than just being able to

send text messages, or make phone calls. Mobile devices

now provide users with a wide range of functions such

as media players, GPS (Global Positioning System) with

advanced computing abilities etc. Many mobile application

are being developed based on context-awareness that would

provide the users with a wide variety of information about

the surroundings such as temperature, noise levels, location

and proximity detected by various environmental sensors.

There are two major problems that can be seen in the

current sleeping habits of people. First, the problem of over-

sleeping due to lack of proper sleep. Secondly, repeatedly

snoozing the alarm clock results in messing up and missing

out on the scheduled work. In our application we provide the

user only with the option of enabling or disabling the alarm

clock application. When enabled, the alarm clock would

automatically set an alarm based on the user’s sleeping

pattern data and the user’s next day schedule. It would

integrate with the calendar application of the mobile and

then sync the user’s events along with their schedule. The

snooze time is not fixed and would get dynamically adjusted

based on the time left for the event considering the location

of the event and the time it takes to reach to the event in

time. After a certain time threshold the application will no

longer provide the user with the snooze option leaving the

user with no other option but to get up from his bed. So”,

our system provides a smart solution for the users to have

an alarm application that sets up an alarm automatically

and provides dynamic snooze patterns depending on various

context-aware information.

2. Background

Context-aware computing is essentially a type of com-

puter operation that anticipates cases of use or, in other

words, works in customized ways based on the context of

user activities. This can apply either to a user’s activities

on the device, or the physical environment in which the

device is being used. By using context-aware computing in

the smart alarm systems, we can develop our system that

cooperates with many sources of information in providing

a harmonious and smart living environment.

In general, there are three main approaches of recom-

mendation systems: collaborative filtering based, content-

based and hybrid methods [1]. Collaborative filtering based

methods focuses on users’ previous activities and predilec-

tions, such as items’ ratings, rather than on content informa-

tion of users or products, while content-based methods uti-

lize item descriptions or user profiles for recommendation.

Hybrid methods, literally, combine collaborative filtering

based and content-based information. As for the smart alarm

system, not only content-based information, but the real-

time situations, such as the traffic or the weather would

affect the choice of snooze time for the alarm. For instance”,

when there is a lot of traffic on the route that the user

has to go from his current location to the destined location

where his event would take place, then the alarm time

and the snooze time also changes with the changing traffic

conditions. Recently, context-aware applications has become

popular in the research community for its good performance

by considering the contextual information[2].

Several researchers have identified and classified the dif-

ferent context types based on their perspective[3][4][5]. One

of the most popular classification was provided by Dey[6]”,

separating context into primary and secondary types. He

stated that with the primary contexts time, identity, location”,

and activity it is possible to fully capture any given situa-

tion. Furthermore, he highlighted that all other contexts are

secondary and can be derived from the primary ones. While

we agree with the importance of the primary contexts, we

do not consider this classification useful. One contradictory

example are machine conditions, such as the temperature

of a machine, which would be considered secondary even

though they cannot be derived from any of the stated primary

contexts. This shows that the distinction between more and

less important contexts is not always possible as it highly

depends on the use case.

When dealing with context, three entities can be distin-

guished : places (rooms, buildings etc.), people (individuals”,

groups) and things(physical objects, computer components

etc.). Each of these entities may be described by vari-

ous attributes which can be classified into four categories:

identity (each entity has a unique identifier), location (an

entitys position, co-location, proximity etc.), status (or ac-

tivity, meaning the intrinsic properties of an entity, e.g.”,

temperature and lightning for a room, processes running

currently on a device etc.) and time(used for time-stamps

to accurately define situation, ordering events etc.).

3. Related Work

Both industry and academia have produced an impres-

sive amount of research work dedicated to providing new

ideas towards building smart applications for a smart living.

While we cannot describe all the applications that exist in

the literature, we summarize prior art that closely relates to

context-aware applications and smart alarm systems.

Many smart applications have been built on context-

based criteria where the applications does some actuation

based on the information sensed from the sensors. Mfundo

Masango et al.[7] proposed an application that provides user

authentication methods that are set up according to the auto-

detection pf areas designated as safe zones by the user. De-

Vaul et al.”,[8] proposed a wearable context-aware reminder

system that delivers reminders based on time, location and

user activity. It focuses on personal context and uses body-

worn sensors to determine in which activity the wearer

is engaged. Cheverest et al.[9], proposes an application

that integrates the use of personal computing technology”,

wireless communication, context-awareness, and adaptive

hypermedia to support the information and navigation needs

of visitors to the city of Lancaster. The context such as

the visitor’s interests, current location, and any refreshment

preferences dynamically affects the interface of the system.

A context-awareness infrastructure in the hospital described

by Bardram and Hansen[10] is aware of the location of

nureses, patients, the trolley and the bed; a bed is equipped

with various RFID sensors that can oidentify a patient

lying in the bed and be aware of patient’s conditions and

treatment.

With extensive research being done in context-aware

applications, the context based models have been integrated

into alarm systems on smart mobiles to provide better

user interaction and smart solutions. Jiaqi Wang et al.[11]

proposed a smart alarm sound recommendation system by

considering not only specific information, such as sleep

patterns, but also context information such as weather, and

social information. Shahreen Kasim et al.”,[12] developed an

Android application that can force the user to wake up by

using the SMS service and the pedometer. The application

needs the user to wake up and walk 10 steps for the alarm

to be disabled and it also sends SMS to the users parents

or friends to wake them up.

4. System Design

Our applications will have only one toggle button to

enable or disable alarm. Once enabled and required permis-

sions are granted (such as calendar access, sleep patterns)”,

the application automatically suggests the alarm time, sound

and snooze frequency. The key components of our system

are raw data collection, feature extraction, data integration”,

context aware model, user history and smart alarm sug-

gestion system. Following subsections explain briefly about

each component.

4.1. Raw Data Collection

To recommend a proper alarm time, sound and snooze

time, it is very important to understand the user. Each

human being has different preferences for these aspects. We

approximate these preferences by analysing his behaviour

such as his recent alarm patterns, selected sound patterns”,

sleep patterns (average sleep timings, etc). Here we also

integrate with other applications such as calendar, weather”,

and location to get the scheduled events for the user and the

current context the user is in. From this raw data, we extract

a set of features as described in the next sub section.

4.2. Feature Extraction

The alarm application is made context aware by ex-

tracting relevant features from the raw data and using these

features to predict what might be the right settings for the

alarm for the current context. Below are the set of features

we propose.

4.2.1. Event Features. For the raw event data that is

collected from calendar app, we extract event features.

Event features consists of next event on the calendar”,

event timings, event location. Suppose if the user has

to attend a conference which needs him to travel 3

hours, it is ideal to set alarm at least 4 hours before

the event. So along with event timings, event location

also is an important feature. Real-time weather also

plays an important role in deciding the alarm time

so we include that as one of the event features.

Figure 1:Proposed System Architecture

4.2.2. User Features. Alarm settings depend on user’s cur-

rent context such as his emotional state, if he is sleeping or

if he is already headed to the event. There are models that

are designed to understand the user emotional state such

as Arousal-Valence model[13]. Using this model, we can

predict if the user is angry, excited, calm or bored. This

prediction can be used to suggest the proper alarm sound.

Each sound in the alarm sound library can be mapped to a

vector to indicate which mood it suits.

4.2.3. User Info. User’s social information such as age”,

gender, occupation, nationality and education level play an

important role in deciding the alarm sound and loudness. If

the user’s occupation requires him to work in a noisy place”,

louder alarms are proper.

4.3. Context Aware Model

Based on the features extracted, user history, our smart

alarm system predicts the alarm time, tone and snooze time.

Each alarm tone in the library is mapped to a feature vector

that can be used to decide which alarm suits for which

environment. These features depend on the loudness, mood”,

tempo, etc of the sound. Context aware model predicts users

current mood, environment (noisy/sleeping/event priority)

and decides which sound to play.

5. Challenges

There will be a lot of complications associated with the

data used in our application. During raw Data collection, the

collected data has to be extremely specific to that user. Our

application requires getting event data from the calendar, it

has to be in sync with the calendar data, any changes in the

calendar application such as creation of an event”,deletion or

updating of an event should reflect the data we have in our

application.

An other form of data source for our application is

pattern data such as alarm”,sleep and selected sound pattern”,

to obtain such data we need to study the user for a long

period of time and use this data and several machine learning

techniques to identify patterns and interests.

We also need to decide what feature extraction models

have to be implemented to come up with features that

give out the maximum information about a user and his

forthcoming events. In the event features, we need to find

how our application will understand the event attributes

such as location, time and make intelligent decisions for

the user to wake him up keeping real time data such as

navigation, scheduling in mind.Our application also has

dependencies with navigation and calendar applications.In

the case of User Features”,Arousal-Valence model has to be

implemented to get his current emotional state. Intelligence

has to be implemented to change the alarm settings based

on user info.

The biggest challenge will be to integrate or keep in

mind user’s calendar data, user’s sleep and sound patterns

and user’s general information and then make a decision.So”,

there will be multiple cases where conflicts might arise.

6. Conclusion

With the implementation of the idea, we proceed towards

developing an application that detects user sleep patterns”,

and the events in the calendar to set the alarm time and

the snoozing pattern of the alarm. With this, the user need

not set up an alarm manually and the application sets up

the alarm for him and also the snoozing patterns are set up

dynamically according to the events in the calendar

Acknowledgments

We would like to thank Dr.Ayan Banerjee for giving

us this opportunity to develop an Idea Paper on Mobile

Computing. We would like to acknowledge the previous au-

thors mentioned in the references for providing a foundation

towards our idea. Lastly, we would like to thank all the team-

members for equally contributing towards this idea.

References

[1] J. Bobadilla, F. Ortega, A. Hernando, A. Gutirrez, Recommender

systems survey, Knowledge-Based Syst. 46 (2013) 109132.

[2] G. Adomavicius, A. Tuzhilin, Context-aware recommender systems”,

in: Recommender Systems Handbook, Springer, 2015, pp. 191226.

[3] Yanwei, S., Guangzhou, Z., & Haitao, P. (2011). Research on the

context model of intelligent interaction system in the internet of

things. In IT in Medicine and Education (ITME), 2011 International

Symposium on (Vol. 2, pp. 379-382). IEEE.

[4] Chen, G., & Kotz, D. (2000). A survey of context-aware mobile

computing research. Technical Report TR2000-381, Dept. of Computer

Science, Dartmouth College.

[5] Chong, S. K., McCauley, I., Loke, S. W., & Krishnaswamy, S. (2007).

Context-aware sensors and data muffling. Context awareness for self-

managing systems (devices, applications and networks) proceeding”,

103-117.

[6] Dey, A. K. (2001). Understanding and using context. Personal and

ubiquitous computing, 5(1), 4-7.

[7] Mfundo Masango, Francois Mouton, Alastair Nottingham and Jabu

Mtsweni(2007). Context Aware Mobile Application for Mobile De-

vices. DOI: 10.1109/ISSA.2016.7802933. Information Security for

South Africa, At Johannesburg, South Africa

[8] DeVaul, R. W., B. Clarkson, and A. Pentland. (2000). The Memory

Glasses: Towards a Wearable Context Aware, Situation-Appropriate

Reminder System, CHI 2000 Workshop on Situated Interaction in

Ubiquitous Computing, Hague, Netherlands.

[9] Cheverst, K., K. Mitchell, and N. Davies. (1998). Design of an Object

Model for a Context Sensitive Tourist GUIDE. Workshop on Interactive

Applications of Mobile Computing (IMC’98), Rostock, Germany, pp.

883-891.

[10] Bardram, J. E. and T. R. Hansen. (2004). The AWARE Architecture:

Supporting Context-Mediated Social Awareness in Mobile Coopera-

tion, Proceedings of the 2004 ACM Conference on Computer Sup-

ported Cooperative Work, Chicago, Illinois, USA, pp. 192-201.

[11] Jiaqi Wang, Yanxiang Guo, Wenhan Han, Jianbo Zheng, Jun Cheng.

Mobile crowdsourcing based context-aware smart alarm sound for

smart living. Pervasive and Mobile Computing – Vol 55 (32-44)

[12] Shahreen Kasim, Hanayanti Hafit, Tan Hua Leong, Rathiah Hashim”,

Husni Ruslai, Kamaruzzaman Jahidin, Mohammad Syafwan Arshad.

SRC: Smart Reminder Clock. IOP COnference Series: Material Sci-

ences and Engineering – 2006.

[13] P. Bustamante, N.L. Celani, M. Perez, O.Q. Montoya, Recognition

and regionalization of emotions in the arousal-valence plane, in: Engi-

neering in Medicine and Biology Society (EMBC), 2015 37th Annual

International Conference of the IEEE, IEEE, 2015, pp. 60426045.