Re: [EToSat] New Version Notification for draft-lhan-problems-requirements-satellite-net-02.txt

[Arashmid Akhavain <[e-mail]>]

Do you mean as an open source initiative? Or more like a formal discussion
in an IETF WG?
I have been looking into LEO/vLEO networks from two point of views:
1- General internet connectivity
2- Integration with mobile networks.

Ad-Hoc networking might be a relevant forum in IETF. We might want to talk
to Alvaro or Erik Kline at some point and seek their opinions when we have
more concrete proposals. IETF Hackathon might be a good venue from a
prototyping point of view.

As for integration with mobile networks, 3GPP started some ongoing work in
this area at the end of its release 16 (I don't quite remember, it could be
release 17) . They do have a long way to go though; as in some of the
proposals they are planning to place a gNB in the satellite. The more
immediate scenario which is also simpler to tackle is RAN to mobile core
connectivity via the LEO network.

Best regards,
Arashmid

On Wed, Feb 23, 2022 at 11:17 AM Raymond Burkholder <[e-mail]>
wrote:

> On 2022-02-23 8:38 a.m., Arashmid Akhavain wrote:
>
> This sounds like an interesting approach. The routing system has to deal
> with both predictable (majority) and unpredictable link events. Traditional
> SDN systems might not be able to perform as the number of nodes goes up
> drastically. Is this work public, or proprietary?
>
>
> Their implementation is proprietary, but many of the concepts are common
> knowledge:
>
>    - based upon TLE information, ground stations can compute satellite
>    feasible neighbors
>    - using network flow algorithms, feasible paths can be computed
>    - there is now enough satellite-local compute available such that
>    multiple ground stations can compute these paths and advertise upwards, and
>    satellites pick the 'best' suggestion for paths to ground (based upon local
>    awareness of rf and link capability/capacity)
>    - the flow calculations take in to consideration the sum of path
>    latencies on the ground, earth-sky, and satellite-satellite
>    - multiple ground stations perform the computation as not all are
>    connected upwards at all times
>
> Is there a 'forum' where some of this could be developed in public?
>
>
> On Tue, Feb 22, 2022 at 10:07 PM Raymond Burkholder <[e-mail]>
> wrote:
>
>> On 2022-02-22 1:27 p.m., Arashmid Akhavain wrote:
>>
>> I agree that existing IP routing protocols can face convergence issues
>> when it comes to satellite networks. IP and existing routing protocols rely
>> on hierarchy to minimise the impact of link events. Establishing hierarchy
>> in satellite networks can be very complex due to various movements
>> especially when it comes to massive constellations.
>>
>> In polar constellations, the movement at the poles, and the seam plane
>> will be problematic for existing routing protocols. The issue of course
>> gets exacerbated in the Walker Delta constellation as the seam effect
>> exists between every other orbit.
>>
>> A new method of routing might be required to address this issue. What are
>> your plans w.r.t to this draft? Have you discussed some of these issues
>> with any of the working groups?
>>
>> Using standard routing protocols are problematic as they are reactive.
>> And due to inter-satellite delays, can take time to re-converge.  I was
>> working with a company to come up with an SDN solution where we used TLE
>> (Two Line Elements) to compute satellite nearness and then pre-populate
>> neighbor entries for installation at pre-arranged times.
>>
>> In addition, to do this properly, these neighbor entries and routing
>> table entries need to be populated based upon distributed path computation
>> capabilities (due to the fact that the same ground stations are not always
>> in view, and paths need to be adjusted as necessary).
>>
>>
>> Best regards,
>> Arashmid
>>
>>
>> _______________________________________________
>> EToSat mailing [e-mail]://www.ietf.org/mailman/listinfo/etosat
>>
>>
>> _______________________________________________
>> EToSat mailing list
>> [e-mail]
>> https://www.ietf.org/mailman/listinfo/etosat
>>
>
>