[Battlemesh] ipv6 prefix delegation in layer 3 mesh networks

[Henning Rogge]

On Tue, Apr 5, 2016 at 4:41 PM, Robert Mader <robert.mader at fu-berlin.de> wrote:
> Hej Henninng,
>
> first of all thanks for being so responsive.
> Well yeah, that was maybe a little bit unclear.
>
> The gateways and node0 are routers/mesh nodes.
> Node 1 and 2 should rather be called clients.
> They're not routers and are connected to node0 for example via wifi.
>
> Between the gateways and node0 could be more mesh nodes, so basically
> this is just the smallest kind of multi-homed mesh network.

The concept of HNCP would be the following.

ALL your mesh routers also run the HNCP daemon. In fact is uses its
own flooding mechanism to distribute the HNCP related data (IP
addresses, prefixes, DNS names).

Both gateway 1 and gateway 2 announce the IPv4/IPv6 prefix of their
uplink as "available" to HNCP.

Each HNCP daemon takes parts of this prefix to configure the Mesh
interfaces with IPs (/32 and /128 ones) and take a small subnet from
each of the distributed prefixes for their attached ethernet, which is
distributed locally via DHCP(v6) and RA.

There is also an optional part that handles DNS names with the help of HNCP.

Henning Rogge