Shangeetha Sivasothy, PhD student, Deakin University & DELH

The PhD research project I am currently undertaking through the Australian Research Council Industrial Transformation Research Hub for Digital Enhanced Living looks into how we can build robust applications for smart homes.

Smart environments already exist in public space, commercial buildings and domestic homes, and are growing in popularity with the rapid development of science and engineering. They are now common and are designed with the contribution of various disciplines such as pervasive and mobile computing, sensor networks, artificial intelligence, robotics, multimedia computing, middleware and software. A ‘smart environment’ is an environment that has the ability to acquire and apply knowledge about the environment and helps its occupants to improve their experience in that environment [1]. Examples of smart environments are smart cities, smart factories and smart homes.

What are smart homes?

Smart homes are the focus of my research project and they, in particular, are being adopted in recent years, with the growth of ageing population. Statista (an online portal for various market data) forecasted that the revenue from the smart home market (eg, sale of smart devices and related services that deliver home automation for private home users) is expected to project an annual growth rate of 16.4%, resulting AUD 2,898 million by 2024, as can be seen in Figure 1. 

Figure 1: Revenue from smart homes worldwide from 2017 to 2024 [11]

In simple terms, a smart home is a residence equipped with the technology that provides assistance and independence to its occupants, more towards helping to provide good health.  Example descriptions of smart homes from the literature are listed below.

Description 1: An application of ubiquitous computing that is able to provide the user with context-aware automated and assistive services in the form of ambient intelligence, remote home control or home automation [2].

Description 2: A residence equipped with technology that allows monitoring of its inhabitants and/or encourages independence and the maintenance of good health [3].

What are example applications in smart homes ?

These smart homes consist “intelligent” applications that are built by software engineers and data scientists. For instance, human activity recognition from motion sensors takes advantage of machine learning models to predict and to understand functional ability and lifestyle of inhabitants [4]. Additionally, there are multiple voice-controlled devices, namely thermostats, refrigerators, light switches and entertainment systems that utilise natural language processing techniques.

A smart refrigerator gives an opportunity to query the contents and various properties of inside contents such as existence, count, category, freshness and size of these contents. In this case, a camera inside the smart refrigerator captures images of these contents and uploads these images to a cloud server. When the smart home occupant raises questions on the contents using a messenger interface, the smart refrigerator replies back with an answer [5].

Another example of such an application within a smart home is a humanoid robot that too exploits natural language processing to aid the cognitive decline in elderly people. An individual’s cognitive functioning naturally gradually deteriorates as a result of diseases, caused by plaques and proteins in the brain. To assess the cognitive decline, the number of syllables per word, sentence length, validity of sentences and word complexity are some metrics proposed as part of the assessment [6].  As human labor for nursing is demanding, humanoid robots that use natural language processing may be used in place to help elderly and disabled people expecting nursing cares as a part of their dependent living in smart homes [7].

What is robustness?

The current study is to look into how robust data science techniques can be identified to the “intelligent” applications that are developed for these smart home systems. Robustness is the ability to tolerate erroneous inputs [8], dependability with respect to erroneous inputs [9] and the ability to function correctly in the presence of disturbances [10]. For example, robustness in smart homes is challenged with malfunctioning sensors or missing inputs from sensors.

Figure 2: Illustration of a smart home technology

What is a use case of robust application in a smart home?

To illustrate the need of a robust smart home, let’s have a look at the following example.

Jane Steward is an elderly lady (aged 80) who lives with her partner in a smart home that is equipped with an instrument X. Instrument X interprets her natural language and can respond to her as if they are having a regular conversation. One day, when she falls off the sofa in her living room due to her prevailing medical conditions, she calls her home nurse through her voice command. However as her voice sounded frail and hence different from her regular voice, instrument X fails to call her registered home nurse in timely manner. This has resulted in Jane Steward being admitted to the neuro trauma unit with brain injury. Instrument X, that was developed to help Jane in emergency situations, had ultimately failed to offer her a helping hand. Hence, the smart home technologies are required to be robust in possible scenarios to avoid failing in such unexpected situations.

What have been done so far?

As part of my PhD project, 57 publicly available data science source code repositories have been examined in detail and 106 data science techniques have been extracted. An initial systematic mapping study has been performed to extract stages of a data science workflow. Combining both, a taxonomy on data science techniques have been constructed and data scientists who contribute towards building “intelligent” applications for smart homes can evaluate these techniques and make informed decisions.

I am hopeful that I will be able to catalogue data science techniques at the end of my research project and will be able to build an interface for data scientists to query the catalogued information. The resulting validated context related references of data science techniques shall be considered universally when selecting robust techniques for developing smart home applications.


Written by:
Shangeetha Sivasothy
ARC Industrial Transformation Research Hub for Digital Enhanced Living PhD scholarship recipient
Applied Artificial Intelligence Institute, Deakin University
NB: The author reserves the right to showcase/publish this blog piece elsewhere and/or in a different medium.

Editorial review by:
Mr David Varley, Partner Investigator
Kevin Hoon, Hub Manager



[1] Cook, D.J. and Das, S.K., 2007. How smart are our environments? An updated look at the state of the art. Pervasive and mobile computing3(2), pp.53-73.

[2] Alam, M.R., Reaz, M.B.I. and Ali, M.A.M., 2012. A review of smart homes—Past, present, and future. IEEE transactions on systems, man, and cybernetics, part C (applications and reviews)42(6), pp.1190-1203.

[3] Chan, M., Campo, E., Estève, D. and Fourniols, J.Y., 2009. Smart homes—current features and future perspectives. Maturitas64(2), pp.90-97.

[4] Hassan, M.M., Uddin, M.Z., Mohamed, A. and Almogren, A., 2018. A robust human activity recognition system using smartphone sensors and deep learning. Future Generation Computer Systems81, pp.307-313.

[5] Gudovskiy, D., Han, G., Yamaguchi, T. and Tsukizawa, S., 2019. Smart Home Appliances: Chat with Your Fridge. arXiv preprint arXiv:1912.09589.

[6] Fredericks, E.M., Bowers, K.M., Price, K.A. and Hariri, R.H., 2018, July. CAL: A smart home environment for monitoring cognitive decline. In 2018 IEEE 38th International Conference on Distributed Computing Systems (ICDCS) (pp. 1500-1506). IEEE.

[7] Bui, H.D. and Chong, N.Y., 2018, September. An integrated approach to human-robot-smart environment interaction interface for ambient assisted living. In 2018 IEEE Workshop on Advanced Robotics and its Social Impacts (ARSO) (pp. 32-37). IEEE.

[8] Abran, A., Moore, J.W., Bourque, P., Dupuis, R. and Tripp, L., 2004. Software engineering body of knowledge. IEEE Computer Society, Angela Burgess.

[9] Avizienis, A., Laprie, J.C. and Randell, B., 2001. Fundamental concepts of dependability (pp. 7-12). University of Newcastle upon Tyne, Computing Science.

[10] Rungger, M. and Tabuada, P., 2015. A notion of robustness for cyber-physical systems. IEEE Transactions on Automatic Control61(8), pp.2108-2123.

[11], retrieved on 06/08/2020

[12], retrieved on 20/05/2020