Edge Computing promises more privacy in the living room

Language assistants control the room temperature, cook coffee and connect with others. A new type of data processing should ensure that privacy is better protected.Edge computing promises more privacy in the smart home.  Source: APSmart home technology at the CES electronics trade fairEdge computing promises more privacy in the smart home.
(Photo: AP)
Munich / Berlin Everything is one, everything is connected: The coffee machine with the light switch with the heating with the door lock - and all this with the Internet. In the House of Tomorrow, a voice command from the residents to let in the bathwater is enough to make a gesture to operate the TV.
When they do what, what they see, what they say and give orders - all of that is valuable information for corporations like Amazon and Google . They use the data and earn billions. With their language assistants Alexa and Co., they come to us as close as never before.
It is therefore hardly surprising that around half of the language assistants who are already in the household today come from these two companies. According to technology analyst Canalys, the total number of such devices worldwide is expected to increase from the current 100 million to more than 300 million by 2022.
Almost three-quarters of them will be equipped with Amazon's or Google's technology, industry experts say. Data protectors are worried about these numbers. As soon as the day-to-day assistants are activated by keyword (eg "Alexa" or "Ok Google"), they transfer unfiltered recordings including background noise into their respective cloud-centralized, often distant server structures with huge storage capacities.
What happens there with the data, the users are no longer with. "Nobody wants everything that happens on site to be transmitted one-to-one anywhere and stored there," warns Thomas Bendig, research coordinator at the Fraunhofer Alliance for Information and Communication Technology in Berlin. It is essential that such information remains with the user.
In the cloud, for example, the data would also be used to improve the machine learning algorithms, "so that Alexa can understand and respond even better," says Amazon on request. And: "We allow our customers to object to data exploitation for training purposes in the Alexa settings."No voice data would be passed on to third parties. In December it came after a report in the magazine "c't" a breakdown: In sent on request files a customer was completely foreign to him records, the group spoke of an "unfortunate event" due to "human error".Computing power on the edge of the cloudThere are already ways to leave users' material on site. One method is called edge computing - computing power at the edge (edge) of the cloud. The data is collected and processed decentrally close to the user.
"This can, for example, take over the normal Internet router that we have in the house," explains Andreas Seitz, a research associate at the Institute for Applied Software Technology at the Technical University of Munich - even if commercial routers have not been able to do so yet.
But it is also possible that such processes run directly on the hardware on which the language assistants are installed. A company that goes this way is the start-up snips based in Paris and New York. Snips manufactures voice assistance systems for business customers who can install them in their devices. The privacy of the users has the highest priority, emphasize the responsible persons.
"It just feels wrong that a corporation has access to a microphone that is in their own home," says Snips co-founder Rand Hindi. The memory and computing power needed to automatically understand the voice commands is therefore not on remote server farms, but on the devices themselves, to which the business customers play the speech recognition of the start-up. Voice and background noise remain in place - without anyone inadvertently accessing it.
More protection and higher speed"There are a number of companies and research projects that are pursuing such approaches," says Bendig. Not only for home use is data processing beyond the cloud attractive. The industry has long since become attentive.
In addition to the protection of information, speed plays a role for them. With intelligent machines, computing processes often have to take place in real time, says Wolfgang Furtner, responsible for edge computing at the Munich-based semiconductor manufacturer Infineon . "The bandwidth and response time of the cloud is limited in the long term, however, if every device in the industry processes its data there."
Much faster it would be if they were charged on site and returned to the machines - especially if the necessary for the cloud Internet connection is not fast enough or incomplete. Edge devices also bundled and anonymized data before going to the cloud. Infineon supplies the hardware for this.Will Edge Computing replace the cloud soon? No, says Seitz from TU Munich. "Edge computing is an extension of the cloud, both worlds are interconnected." Especially in artificial intelligence or machine learning, you need huge computing power to process enormous amounts of data. "Doing these things in an Edge and learning that is unrealistic," says Seitz.
Filtering, bundling and sending only the bare necessities to the cloud - that's what Edge devices can do, experts agree. For privacy, that would be an advance.