Re: [6tisch] 6tisch join requirements for 6top

[Michael Richardson <[e-mail]>]

{can I subscribe Tero to the 6tisch-security mailing list?}

Tero Kivinen <[e-mail]> wrote:
    >> > Note, that receiving peer needs to know the extended address of the
    >> > sender in the 15.4 to be able to decrypt the packet, as the nonce
    >> > generation depends on that. There is no way around that. 
    >> 
    >> When you say "extended address", do you mean 64-bit EUI?

    > Yes. In 802.15.4 the extended address is the full 64-bit EUI. The
    > short address is the 16-bit address assigned by the coordinator (and
    > that short address is associated with PANId). 

    > The 15.4 uses nonce for the AES-CCM in format of

    > +------------------------+------------------+--------------------+
    > | Octets: 8     	 |	 4	    |	    1            |
    > +------------------------+------------------+--------------------+
    > | Source Address	 |   Frame Counter  |  Security Level    |
    > +------------------------+------------------+--------------------+

    > for non-TSCH mode, and for TSCH mode it uses:

    > +------------------------+-------------+
    > | Octets: 8     	 |     5       |
    > +------------------------+-------------+
    > | Source Address	 |     ASN     |
    > +------------------------+-------------+

    > where the ASN is the absolute slot number. The Source Address is the
    > 64-bit EUI, i.e. extended address of the sender.

    > The source address is fetched from the KeyDescriptor found using the
    > Key Identifier Mode, Key Source and Key Index.

    > So unless you know the extended address of the sender, you cannot
    > decrypt the packet, as you cannot calculate the Nonce for the AES-CCM
    > operation. 

What I'm reading is that multicasted messages must always use extended
addresses if they are secured.   For other messages, if one has a node to node 
security, then one can map the 16-bit address to an extended address, and
create the nonce.  If one does not have node to node security, then the
receiving node might not have the mapping: does that mean one has to use
extended addressing whenever there is a single, network-wide key?

-- 
Michael Richardson <[e-mail]>, Sandelman Software Works
 -= IPv6 IoT consulting =-