Ockam logo
Product
OverviewCloud SDKEdge SDKEmbedded SDKRegistryRouter
Contact us

Secure Channels

Secure Channels provide end-to-end secure and private communication that is safe against eavesdropping, tampering, and forgery of messages en-route.

Machines, within IoT systems, communicate amongst each other by exchanging messages over a variety of transport protocols like TCP, Bluetooth, ZigBee, LoRaWAN etc. Many of these protocols also come with some means of establishing secure communication, however, the security properties of these mechanisms can vary in many subtle but significant ways.

In case of TCP, communication is secured using TLS which is the most mature amongst the various secure transport protocols that are commonly used within IoT systems. While the latest version, TLS 1.3, is notably simpler than previous versions, TLS remains intrinsically complex in order to support the needs of legacy web browsers and web servers.

These legacy requirements don't usually apply to IoT systems and TLS' complex negotiation logic is an overhead. A variety of configuration options and more cryptographic agility, than what is required, also make TLS challenging to configure correctly in large scale IoT deployments.

Furthermore, in many IoT scenarios, a message travels over multiple transport protocol connections before reaching its destination. For instance, a message from a heart rate monitor to a phone application would typically require at least two transport connections. The monitor makes a TCP connection to a server on the Internet, the phone also makes a different TCP connection to the same server.

Heart Rate Monitoring

This is typical because usually neither device is listening, as a TCP server, at a fixed public address. Instead, both devices connect, as TCP clients, to a well know server address and that server then routes messages from the monitor to the phone and vice versa. This topology can handle various routing related complexities like when either device is roaming and their IP address is continuously changing or when either device is offline or if either is behind a NAT.

Since TLS is usually tied to a TCP connection, this means that even in cases where TLS is configured correctly our data is entirely exposed to the server whose primary task, in our topology, was supposed to be routing and caching of messages. All protections provided by TLS against eavesdropping, tampering, and forgery of messages are lost at the server even though routing and caching don't even require it to know the entire contents of a message.

This violates the security design principle of least privilege

Heart Rate Monitoring with TLS

Ockam includes an end-to-end secure channel protocol that can span across multiple transport connections.

Heart Rate Monitoring with Ockam

This secure channel protocol can be layered on top of most common IoT transport protocols to provide a consistent, simple to configure end-to-end secure channel even in scenarios where multiple transport protocols are in the path of one message.

Flood Monitoring with Ockam

Many secure communication protocols, other than TLS, that are commonly used within IoT systems tend to be lacking in their design and security properties. Usually there is no forward secrecy, no protection against key compromise impersonation attacks and generally poor hiding of identities.

Flood Monitoring with Ockam

Decoupling the secure channel protocol from the transport protocols removes the complexity of dealing with multiple disparate secure protocols, enables us to have end-to-end secure channels, minimizes the attack surface and helps us to build better end-to-end secure and private systems.

The precise design of the secure channel protocol is being discussed as part of Ockam Proposal OP-0002.

Previous

How to build a vault implementation

Next

Transports

Edit this page
Star Ockam repo