The popularity of smart homes is on the rise but with it comes increased concerns about security. Now an article in CBR online.com asks if machine learning will be the answer.
It’s expected that by 2020 there will be 40.9 billion active wireless connected devices, largely consisting of non-herb devices such as sensors. By the same year, is also been predicted that the market for the Internet of Things will be worth in excess of $7 trillion worldwide. There is increasing concern over the safety of personal data that is uploaded into the cloud via these systems.
It is already accepted as a given that personal data is being used for legitimate purposes, but the worry is that as we transition closer to always being online, data will be exploited. Two years ago, it was discovered that it was possible to gain access to smart hubs and to open locks or reconfigure alarm systems and just last year it was discovered that household names including Samsung and Google had serious flaws in receiving data transmitted from smart homes, leaving this data vulnerable to being misused.
The personal nature of this information has possibly quite sinister implications, allowing a third party to determine when a person is not at home. Even when the person is at home, this personal data makes it easy to ascertain exactly what they are doing at that moment. One emerging trend is the use of botnets, systems in which large numbers of botnets can combine their power to initiate attacks that may potentially shut down websites. The reason why these could be dangerous is their ability to control sensors and remote cameras in a house.
The article raises the questions as to how these systems could be secured. One possibility is to use machine learning. Data is harvested through platforms such as Avast and AVG and which are used by around 400 million people worldwide. This data can then be input to discover the most effective ways to reduce these problems. Being able to access such large sample sizes will allow companies to help protect users through improved security software. The problem is that many devices found in smart homes do not have the storage and processing capacity to run these types of programs.
It’s possible that machine learning could reduce the requirement for these systems through observing data patterns from a central hub. It offers the possibility of being able to use data to protect us rather than this information being used against us.
