This method is what Signal currently does: group messages are sent through pairwiseĬhannels in a way that actually makes them indistinguishable from private Users in a group, the cost of sending a message is then in O(N). The most straightforward way to obtain a group is to simply createġ-on-1 channels for each pair of participants. We first have a look at what messaging applications currently do, then we'll see howīinary trees could optimize the protocol(s). How impractical it is to scale-up from 1-on-1 to 1-on-Many without making compromises. Some with remixes or their own idea, but all of them presenting the same issue of Messaging applications, some of them with the same basis (the Signal protocol, usually), There are a lot of different so-called secure

the key(s) and the group of people can change over time. Group messages: two people are already a group!įormally, the protocol proposed by MLS is referred to as a Continuous Group KeyĪgreement, in the sense that several people agree on a shared key continuously, Without really making a difference between private messages and MLS on the other hand wishes to immediately propose a solutionįor conversations involving two or more members, and aim to scale to groups with tens of thousands of members, Most messaging protocols focus on end-to-end encryption for 1-on-1 conversationsįirst, and later extend that protocol to the group setting. This effort is divided in two different documents specifying: (iterative, following the IETF procedures, open to suggestions and contributions from anyone). Messaging protocol, similarly to what is done with TLS (hence the name),īoth in terms of security properties and the process to define the specification Was created with the idea to standardize the security layer of a In 2018, the Messaging Layer Security (MLS) working group

Signal group chat encryption