Marlin's TEE network

DeepWorm uses the Marlin network to run the Nematoduino simulation verifiably

Marlin provides an easy to use framework to wrap applications in docker containers and easily deploy them in Trusted Execution Environments (TEEs). This ensures that applications run autonomously and that their execution can't be manipulated by external parties including the host machine's administrator. This is important as DeepWorm seeks to produce responses to stimuli just as a real worm's nervous sytem would which necessitates that no individual be able to manipulate the outputs of the Nematoduino simulation.

Trusted Execution Environments (TEEs)

TEEs are a protected area of memory and CPU where data and code are isolated from other processes usually at a hardware level or through encryption. In this way, they ensure the confidentiality of data and integrity of computation running in them as not even the kernel can read or tamper with code executing in them. This property is especially useful in decentralized networks where data and computation might be leaked or manipulated by untrusted hosts in whose machines the data is stored and processed.

DeepWorm uses Marlin for the following properties:

• Reproducible builds: It is necessary that users are able to ensure that the executable that they are interacting with was built using the same open-source code that's been vetted by them and others

• Remote attestations: Anyone can build and deploy the enclave hosting the worm's neural network simulation. All they need to do is to get their enclave attestation verified onchain to prove that they are running a genuine enclave. The worm can therefore live forever and be easily cloned.

• Cross-chain support: It should be possible to verify attestations across L1s and L2s to track its movement across any chain.