What is an Oracle or Middleware?

Oracles are information bridges (known as middleware) between blockchains (On-Chain) and the real world (Off-Chain). They send and verify information relevant to a blockchain or Smart Contract via data providers, web APIs, or IoT devices. The main reason for their existence is because blockchains are deterministic and are simply limited to obtaining information that is already inside the network itself and not outside it. Smart Contracts on the blockchain will be executed based on real-world inputs and outputs. A hybrid Smart Contract is an oracle smart contract that receives data updates via API queries.

• Hardware Oracle

They are designed to interact with the physical world and obtain information from it to feed into Smart Contracts. An example for the transmission of such information could be through sensors or RFID systems converting real-world values into values understandable to smart contracts.

• Software Oracle

They provide real-time information and that makes them one of the most common for DeFi or Dapps cases, such as gambling. The oracle extracts data via APIs to third-party information sources and sends it to the smart contract on the blockchain.

The problem of the oracle

Oracles are a very important part of the blockchain ecosystem because smart contracts will execute decisions based on the data provided by the oracle. Therefore, if the oracle is compromised, the smart contract will be compromised as well, which is known as the "oracle problem". As a general rule, anything that is centralized is more susceptible to attack. In the case of centralized oracles, they are more prone to man-in-the-middle attacks where data is intercepted between the blockchain and the oracles to modify that data. This is why centralized oracles are not considered secure.

Decentralized oracles

Decentralized oracles (DON) together with the creation of hybrid Smart Contracts obtain information from multiple sources and in a decentralized way. The result is communicated to the Smart Contract through a consensus where the majority of sources are accepted, thus providing a higher degree of trust. The use of decentralized oracle networks is a next step in the adoption of Blockchains and Smart Contracts through the corroboration of Off-Chain data. This maintains decentralization at all times, confidentiality, high scalability and high reliability. For example, in a simple multi-oracle network the approval of the veracity of the data could be divided into multiple parts where to verify the data, an O_i oracle needs to compare the data received x from a sensor with the data obtained y by the oracle giving a value 0/1 depending on the result (true if they match, false if they do not).

ADOS (AirTrace Decentralized Oracle System)

ADOS is primarily aimed at capturing anomalies between data collected by different sensors and providing a degree of reliability of useful data. By using Artificial Intelligence to provide a Data Quality Factor (DQF) and a distributed oracle system on the iExec platform, ontological relationships between sensors in the market can be leveraged to detect anomalies in the readings collected from IoT sensors.

The IoT devices generate the collected data and send it to the Airtrace platform on a regular basis. The data will be sent to the selected Blockchain as set by the customers (e.g. for 24 hours by accumulating all the readings of the day and including them in a single Blockchain transaction). In this case, before injecting into the blockchain, AirTrace sends the accumulated set of IoT readings to ADOS where the readings will first be shared with the iExec Doracle. system. Using a decentralized anomaly detection model with a set of "workers", the input data is processed and returned to ADOS along with a data quality factor (DQF).

Each independent worker will maintain a copy of the docker application running in a TEE (trusted execution environment) and take the model weights and metadata from IPFS/ S3 using a consensus algorithm configured on the iExec platform. This data will be forwarded to AirTrace to issue a new blockchain transaction that includes the original data and the DQF (along with the hash, anomaly detection model information, consensus algorithm used and other metadata). Finally, ADOS periodically updates the model weights and the necessary metadata, depending on the continuous training of the AI model and its subsequent storage in IPFS and/or S3 where the model parameters are acquired by the set of "workers".

All this information can be very relevant for possible subsequent audits, in addition to all the benefits discussed in the previous article. AirTrace also aims to make the code open source for the ONTOCHAIN community.