DISCLAIMER: I am not a patent attorney, and I am not well versed in GPON networks. The following is my layman's take on US Patent 10,904,649.
I took a look at the patent that you referenced (US Patent 10,904,649).
After reading the text and looking at the drawings, I agree with your conclusion that the invention disclosed in the '649 patent generally aims to solve the problem where peers on the same GPON can see each others' traffic. However, I don't believe the method claimed in the '649 patent uses encryption to solve the problem.
As you point out, methods to solve this problem have existed for a many years (long before the application was filed in 2019). The fundamental requirements for patentability are uniqueness and non-obviousness. So, what's unique and non-obvious here?
To answer these questions, I took a look at the claims of the '649 patent, to see exactly what is claimed in the specification. As you may know, the claims of a patent specify exactly what systems/methods another entity may not [build | use | sell | export] - without obtaining a license to do so from the assignee of the patent (otherwise this entity would be infringing the patent).
There are only four claims in the '649 patent, and only one of these claims (claim 1) is an independent claim (the other four claims are dependent claims of claim 1, which means they are 'special cases' of claim 1). Claim 1 reads:
1. A digital data network communication method comprises:
accepting a plurality of private data streams into at least one passive optical network interface router serving a plurality of private user devices through a plurality of optical network units (ONUs);
configuring said at least one passive optical network interface router to virtually separate information intended for each of said plurality of private user devices, wherein said configuring comprises:
generating an independently unique virtual routing table for each of said plurality of private user devices using Virtual Routing and Forwarding (VRF);
uniquely tagging ethernet frames contained in said plurality of private data streams and intended for said first one of said plurality of private user devices according to entries in a first one of said virtual routing tables associated with said first one of said private user devices using Multi Protocol Label Switching (MPLS) to form MPLS tagged data packages;
sending said MPLS tagged data packages to a passive optical networks optical line terminal (OLT);
aggregating within said OLT said MPLS tagged data packages into a common data feed; distributing said common data feed to said plurality of ONUs;
wherein said distributing comprises: replicating said common data feed using at least one optical splitter connected to said plurality of ONUs;
extracting, at each of said plurality of ONUs, from said common data feed those MPLS tagged data packages intended for a specific one of said plurality of private user devices.
As you can see from the above claim - there is no mention of encryption, or anything relating to encryption, anywhere in the text. However, after reading the above claim, my feeling is that this a very specific method of solving the problem where peers on the same GPON can see each others' traffic.
Often with patent prosecution, the original patent application filed by the inventor contains very broad claims for systems/methods aimed at solving some problem. During examination, the examiner often finds 'prior art' that predates the systems and methods described in the claims. The inventor can then go back, and narrow the scope of their claims, or eliminate some of the claims altogether, to avoid 'reading on' the prior art. Then, the inventor can submit the revised claims for re-examination. This tends to result in fewer claims that are much narrower, and much more specific. This may have been what happened with patent application that eventually matured into the '649 patent. A look at the patent file wrapper (which shows the back-and-forth between the inventor's attorney and the patent examiner during examination, as well as each revision of the inventor's claims) would show if this is what happened in this case.