[manet] Re: MANET Internetworking: Problem Statement and G ap Analysis

Hello AB, and thank you for these comments. Please see below for responses:

Regards – Fred Templin

From: Abdussalam Baryun <[email protected]>
Sent: Thursday, September 25, 2025 10:13 AM
To: Templin (US), Fred L <[email protected]>
Cc: [email protected] List <[email protected]>
Subject: Re: [manet] MANET Internetworking: Problem Statement and Gap Analysis

Hi Fred,

I read the draft but still need to see the interest/solution in disaster_scenarios, which your assumption was not clear to me, as we using 5G network/cells or only MANETs, I think your use case is using the internet of wireless_terrestrial_network/5G_network, so I don't think we can say MANET of MANETs, because the 5G network already is using its management protocols and routing protocols that are different to MANET. Or please explain from where some of the cell phones/router are getting the internet service? I think the draft needs to identify the terrestrial network as use cases, IMHO to make it general will never help.

>> Yes, the expectation is that Internet-facing proxies will receive Internet access from provider infrastructure
>> through traditional service provider models such as 5G, then provide MANET-facing interfaces to support
>> Internet connection-sharing services to MANET nodes that cannot directly engage provider infrastructure
>> themselves. So, maybe a more acceptable way to phrase the big-picture view of an Internet structured in
>> this way is a “network-of-(Mobile Ad-hoc) networks”?

AB> more below

On Wed, Sep 3, 2025 at 11:54 PM Templin (US), Fred L <[email protected]<mailto:[email protected]>> wrote:
See below for a new draft motivated by recent discussions on the MANET mailing list
and charter discussions at the IETF 123 MANET working group session. The draft
identifies two distinct types of MANETs:

1) True stub MANETS - those for which an Internet connection sharing node acts
as a regular IP router on behalf of a downstream-attached personal local network
consisting of end systems that can provide internal multihop forwarding services
within the local area. An example is a cellphone with an attached 802.11 network
operating in "ad-hoc" mode where some nodes may be separated by multiple WiFi
hops. Such MANETs are likely to include a small-to-modest number of nodes in a
common MANET routing protocol instance, and the local MANET nodes should be
able to use (pseudo)random addresses taken from a suitably large address
assignment space with little/no risk of collision.

yes this is ok

>> OK.

2) Not-so-stubby MANETs - those for which one or more Internet connection sharing
nodes act as proxys at a layer below IP on behalf of peer nodes that may be multiple
hops away from the nearest Internet infrastructure element. An example is a disaster
relief team that enters an area of degraded infrastructure and where each person's
cellphone supports device-to-device multihopping over the 5G/6G interface itself
while providing Internet connection sharing services based on opportunistic
infrastructure contacts. Again, such MANETs are likely to include a small-to-modest
number of nodes engaged in a common MANET routing protocol instance.

Ok, we have more than one MANET, so we will have let say three MANETs with established addresses.

>> I think it is still useful to consider such network as a single MANET but with multiple
>> candidate Internet connection sharing proxies. The MANET nodes can travel together
>> limited to doing intra-MANET multihopping unless or until one or more of the Internet
>> connection sharing proxies comes in contact with Internet infrastructure. Each such
>> proxy can then extend Internet services to other nodes in the MANET, but the whole
>> thing can still be considered as one MANET.

>From an Internetworking standpoint, however, each cellphone that connects
to the Internet should be considered as a potential MANET router where the
Internet itself becomes a "MANET-of-MANETs".

The internet provider's net should not be considered as MANET, usually that internet is the wireless infrastructure_terrestrial_network, and not MANET of MANETs. Usually our MANET is at the layer 3 or 4.

>> Agreed we can drop the confusing terminology so as to not imply the global Internet
>> has anything to do with MANET. However, the stub and not-so-stubby MANETS that
>> connect using the global Internet as transit can be implemented as an adaptation
>> layer abstraction, i.e., neither layer 3 nor 4.

The MANET Internetworking
service should then support interconnection of MANETs using the Internet as
transit based on virtual circuits established on-demand as an adaptation
layer function.

ok so do you mean that is at layer 4 ? we need to be clear to determine the service access point.

>> No, I mean as an adaptation layer service manifested through encapsulation. The
>> adaptation layer appears below the network layer but above the data link layer in
>> the architecture.


So, the Internet would serve as a massive transit network for connecting
small-to-modest sized MANETs on demand. With billions of MANET routers
in the Internet, this means that the likelihood for address collision becomes
non-negligible as demand-driven MANET--to-MANET virtual circuits are
established and dissolved. This means that MANETs that opportunistically
connect to the Internet require some form of administratively-assigned
adaptation layer address - the Multilink Local Address (MLA).

usually this adaptation layer address is already provided by the internet provider's net, so do you mean that your scenario assumes that there is no administrative assignment provided within that large network that covers disaster locations/regions?

>> In considering MANETs that need to operate when Internet infrastructure is either unavailable or
>> only intermittently available, there is a need for an address type that is not provider assigned; it
>> is an address type that each node procures independently from a registration service for itself
>> with assurance that no duplication occurs. It is similar to a MAC address, but it is a fully-qualified
>> IPv6 address and not a MAC address at all. It is called  a Multilink Local Address (MLA).

we need to answer the above before considering the number of mobiles per region and number of MANETs per disaster solution.


So, the concern is not for massively-large MANETs that operate in a
common routing protocol instance - that never was and never will be
a practical environment for MANET routing protocols.

ok, I understand

The concern is
for vast numbers of small to modest-sized MANETs that interconnect
using the Internet as transit where there are no constraints on the
nature of MANET-to-MANET interconnection events.

I think the draft needs more definitions and clarifying the disaster networking.

>> Would it help to include a section on use cases? I think others have tried to do that in the past
>> but did not receive traction. If you think use case text would be helpful, I think it should be a
>> couple of paragraphs at most and not multiple pages otherwise the draft could become
>> marginalized.

Best Regards
AB

_______________________________________________
manet mailing list -- [email protected]
To unsubscribe send an email to [email protected]