An IoT data sharing dilemma: Transparency or Translucency?
A blockchain solution for transactive grid
Nowadays, Internet-of-Things (IoT) is gaining momentum as the technology of the future. The main goal of IoT, like the name may suggest, is to bring the internet on all kind of devices. Future generation of industry, automotive, house, energy and other fields will be composed by smart devices, i.e. connected to the internet to interact, with some extent, with other devices. Currently, a number of smart-* products are already available in everyday life. The most prominent and common devices are smartphone. Their success mostly is due to the portability; indeed, everybody brings a smartphone with herself in all the moment of the day. The internet connectivity is the key feature which allows smartphone to be an essential item for the user. News, mobility, shopping and other kind of services have been improved and simplified, giving to the user a higher quality of experience. Other smart devices are arising like smart tv, smart camera, smart watch and so on, all of them providing communication capability.
The IoT paradigm aims to extend such a connectivity with all kind of devices. We will get more and more used to devices with the “smart” prefix. Smart-industry, smart-home, smart-grid, smart-car and many others in few years will be actual solutions to a number of different fields.
An IoT environment should be independent and able to change itself automatically. Indeed, through Artificial Intelligence algorithms they can continuously learn from stored data to predict their future behaviour and automatically trigger operations. For example, in a smart home environment, the system can collect data related to the user behaviour during the day in order to autonomically anticipate to turn on the heating according to the forecasted moment the user wakes up or comes back home. In the same way it can anticipate to turn it off when he is going to go out.
The accuracy of Artificial Intelligence algorithms is strictly related to the amount of data used to train them. Indeed, inspired by the human neural network, they are mathematical structures which learns from data without neither an a priori knowledge nor a predefined model. Hence, the more data you pass them the more they learn and predict correctly.
To improve the accuracy of such algorithms the research community focuses on data sharing. Although, data sharing allows different systems to have a more accurate global knowledge, the main issue to cope with is related to the security. In the smart home environment, for example, sharing with the others your attitude, like when you go out and come back home, may bring both a privacy and a security issue. In the automotive environment, similarly, sharing among cars the position can simplify the algorithms of GPS navigators to select the best way to reduce the overall traffic congestion, but it also may affect your privacy.
It is clear how data sharing must be so regulated, especially when data are sensitives like for example patience data stored by hospitals.
Yet the main question is: how to regulate such data sharing?
Two main philosophies can be considered: transparency and translucency. Such definitions took inspiration by the physical properties of the light. The former allows light to pass through the material without being scattered, conversely the latter allows only a subset of light to pass. The metaphor with data can be seen as who can access data. In a transparent approach everybody can see everything, while with a translucent approach only a subset of allowed users can see data while they are obfuscated to others.
The first is gaining momentum especially in the public sector as everybody can verify the government actions. Citizens, specifically, can use public data to trace such actions and trust them. Conversely, translucency, hinders users to know other data and so it is considered a more effective approach in a peer-to-peer (p2p) network.
Anyway, it is not always easy to clearly map a scenario in one or the other. For example, in a p2p network of hospitals sharing citizens’ data, there should be transparency or translucency? If you trust to all hospitals involved the answer can be transparency, but what if a hospital is subverted with a cyberattack to steal or tamper with such data?
PETRAS is a UK hub to incubate projects related to cybersecurity of IoT. During the last meeting of November 2017, researchers involved in the PETRAS community share their ideas as well as the main problems to address. Specifically, one of the main topic has been how to define a general model for data sharing by trading off transparency and translucency.
During the last years, blockchain came out as a disruptive technology for cybersecurity. Nodes joining a p2p network share a replicated ledger containing an ordered list of transactions. Tampering with data is very hard as the attacker needs to subvert the majority of the nodes (or the computational power of the network). Blockchain success mainly derived from provided properties of decentralisation and data integrity. Since each node stores the entire list of transactions issued in the network, blockchain is a robust solution to provide transparency as everybody can check all transactions issued. Thanks to smart contract, i.e. programs deployed and executed on top of a blockchain, a wide range of application has been assessed from supply chain to secure storage. A new perspective is the integration of such decentralised technology within the IoT world.
Along such a direction, in the PETRAS hub we proposed BlockIT, a project aimed at integrating a transparent blockchain for a privacy preserving translucent data sharing on a smart IoT-aided grid. More specifically we propose an architecture for a secure new generation transactive grid where a set of prosumers may trade energy each other without a trusted third party.
Through smart contracts, there are regulations established for the interaction between the nodes. Using a blockchain we have a transparent way to handle the transactions and by employing an effective wallet of pseudonyms we propose a solution to also preserve the privacy of prosumers data.
Thus, a prosumer cannot figure out a specific behaviour of another prosumer, but it can verify the correctness of each transaction.
In that way, data related to energy usage can be shared to predict the amount of energy needed by the microgrid of prosumers. Furthermore, an attack to some nodes to execute an invalid transaction (e.g. trying to sell an amount of energy that you do not have) fails as all transactions have to be agreed with the consensus of the network, as required by the blockchain protocol.
Following an eprint of the work submitted to the PETRAS conference:
Can the proposed privacy preserving blockchain be a solution to accommodate both transparency and translucency?
