LoRaWAN packet routing (HIP draft) #9

Vagabond · 2020-02-14T02:07:54Z

Rendered view: https://github.com/helium/HIP/blob/c2f3ce61466b003731bb967959ca8b6e7706fca5/0008-lorawan-routing.md

lthiery

This seems like a very nice solution to the problem of on-boarding devices with arbitrary credentials. While I believe we should focus on this routing scheme in the short-term, I would note that we have two paths to reduce Join latencies should that become a friction point for network usage:

we could use a reserved DevEUI range to be a routing "routing mode" indicator; this would allow you to use part of the DevEUI or the AppEUI as a way to signal OUI/deviceID
we could use the MType bit field to make a whole different message, either Join or normal message

lthiery · 2020-02-19T17:05:49Z

0008-lorawan-routing.md

+# Motivation
+[motivation]: #motivation
+
+#Why are we doing this? What use cases does it support? What problems does it solve? What is the expected outcome?


delete prompt perhaps?

0008-lorawan-routing.md

amirhaleem · 2020-02-20T19:21:04Z

This seems like a very nice solution to the problem of on-boarding devices with arbitrary credentials. While I believe we should focus on this routing scheme in the short-term, I would note that we have two paths to reduce Join latencies should that become a friction point for network usage:

we could use a reserved DevEUI range to be a routing "routing mode" indicator; this would allow you to use part of the DevEUI or the AppEUI as a way to signal OUI/deviceID

we could use the MType bit field to make a whole different message, either Join or normal message

We should be cognizant that no off-the-shelf LoRaWAN devices will work this way - we ideally want a solution that is compatible with any LoRaWAN device, not requiring any changes to their credentials or firmware. I'd argue overloading DevEUI is not much of an improvement over the current AppEUI hack we are using today.

amirhaleem · 2020-02-20T19:28:18Z

0008-lorawan-routing.md

+
+## What parts of the design do you expect to resolve through the HIP process before this gets merged?
+
+We need to nail down the permissible sizes of the XOR filters we want to store.


I think you will have to propose something here @Vagabond, I am not sure how to evaluate this.

amirhaleem · 2020-02-20T19:29:01Z

0008-lorawan-routing.md

+We need to nail down the permissible sizes of the XOR filters we want to store.
+
+We need to decide if these routing table entries also contain a range of
+DevAddr entries to be used for routing of non-join packets.


What are the tradeoffs here? If we don't pre-assign DevAddr's to OUI's, how would a Hotspot lookup the correct OUI on subsequent non-join uplinks?

I have added a second HIP to this PR that addresses DevAddr routing more thoroughly

amirhaleem · 2020-02-20T19:29:46Z

0008-lorawan-routing.md

+
+## What related issues do you consider out of scope for this HIP that could be addressed in the future independently of the solution that comes out of this HIP?
+
+Pricing and availability of routing table entries for organizations operating


The economics are out of scope here - will kick off a separate discussion/HIP on this.

amirhaleem · 2020-02-20T19:31:54Z

0008-lorawan-routing.md

+
+## What parts of the design do you expect to resolve through the implementation of this feature?
+
+We need to define where/how these routing entries are stored, how they're


Similar to earlier comment I think you should propose a solution (or just start building for future review) - I'm assuming you are talking at the ledger/rocksdb level, and what the transaction structure should look like?

lthiery · 2020-02-20T19:52:24Z

We should be cognizant that no off-the-shelf LoRaWAN devices will work this way - we ideally want a solution that is compatible with any LoRaWAN device, not requiring any changes to their credentials or firmware. I'd argue overloading DevEUI is not much of an improvement over the current AppEUI hack we are using today.

Right - the proposed solution here is great because it allows for arbitrary credentials and creates universal compatibility for off-the-shelf LoRaWAN devices.

I'm just leaving notes on a few ideas on how to reduce latencies while remaining compatible with this proposed scheme; I am not suggesting either of those routes be considered for development today.

Co-Authored-By: Louis Thiery <louis@helium.com>

lthiery · 2020-02-21T05:47:15Z

0008-lorawan-join-routing.md

+the command in the README. The benchmark was first run on the Intel machine to
+generate the routing tables, run a second time to use the generated routing
+tables, and then the tables were copied to the Pi to run the test with the same
+databases (generating the tables can be quite RAM intensive, so it is unsuitable


How will tables be updated then? Every time someone registers a set of IDs, someone has to do the work to generate and distribute the new table?

Yes, so you'd likely want to batch updates.

0009-lorawan-devaddr-routing.md

evanmcc · 2020-02-22T00:08:56Z

0008-lorawan-join-routing.md

+XOR filter. Both the DevEUI and the AppEUI (totaling 128 bits together) would be
+hashed to a 64 bit key, this should help in cases of DevEUI collisions.
+
+A new ledger entity would be added to the blockchain, a 'routing entry'. These


The language in the next few paragraphs is unclear. Is there one xorfilter per OUI or one big one?

At least one per OUI, for very large OUIs there might be several (so we limit the size of any single object we have to examine).

evanmcc · 2020-02-22T00:11:26Z

0008-lorawan-join-routing.md

+tables, and then the tables were copied to the Pi to run the test with the same
+databases (generating the tables can be quite RAM intensive, so it is unsuitable
+to be done on the Pi).
+


I'd love to see some extension briefly discussing the process of generating the table or tables. Who do we trust to do this? How do we certify the validity of a table? How do we keep a customer from shooting themselves in the foot?

Added something

evanmcc · 2020-02-22T00:24:21Z

0008-lorawan-join-routing.md

+milliseconds on an raspberry pi 3b+ in 32 bit mode. The smallest xor filter can
+be checked in well under a millisecond.
+
+In a more complete test, 10,000 OUIs holding device space for 50 million devices


Are we doing anything clever here, or just checking them one by one?

We have to do a linear scan, as far as I know.

0009-lorawan-devaddr-routing.md

PharkMillups · 2020-06-12T12:17:50Z

This is mostly done. @Vagabond is going to give it one more review then we'll merge.

jamiew · 2020-12-22T14:37:26Z

In the interest of keeping PRs moving, I am closing this as stale. Please re-open if you'd like to finalize and merge it for discussion.

Vagabond added 5 commits February 13, 2020 18:07

WIP

156a98f

Update

ffdb784

Fix subsections

0273838

Fix

c2f3ce6

Fix

48a8b8f

lthiery reviewed Feb 20, 2020

View reviewed changes

amirhaleem reviewed Feb 20, 2020

View reviewed changes

Vagabond and others added 3 commits February 20, 2020 15:32

Update 0008-lorawan-routing.md

04bdd36

Co-Authored-By: Louis Thiery <louis@helium.com>

Break DevAddr routing into a seperate HIP

002f9de

Add more deets

e5f6096

lthiery reviewed Feb 21, 2020

View reviewed changes

vihu reviewed Feb 21, 2020

View reviewed changes

0009-lorawan-devaddr-routing.md Outdated Show resolved Hide resolved

Updates

7385035

evanmcc reviewed Feb 22, 2020

View reviewed changes

Review comments

5b7430c

evanmcc reviewed Feb 22, 2020

View reviewed changes

vihu reviewed Feb 22, 2020

View reviewed changes

0009-lorawan-devaddr-routing.md Outdated Show resolved Hide resolved

Vagabond and others added 2 commits February 21, 2020 16:37

Less average

544d8fd

update from range to subnet

8ae86f7

jamiew added the draft label Jul 3, 2020

jamiew changed the title ~~LoRaWAN packet routing~~ LoRaWAN packet routing (HIP8 draft) Aug 26, 2020

jamiew added the stale Old and needs attention label Nov 3, 2020

jamiew changed the title ~~LoRaWAN packet routing (HIP8 draft)~~ LoRaWAN packet routing (HIP draft) Nov 4, 2020

jamiew closed this Dec 22, 2020

jamiew deleted the adt/lorawan-routing branch June 30, 2021 14:36

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

LoRaWAN packet routing (HIP draft) #9

LoRaWAN packet routing (HIP draft) #9

Vagabond commented Feb 14, 2020 •

edited by amirhaleem

Loading

lthiery left a comment

lthiery Feb 19, 2020

amirhaleem commented Feb 20, 2020

amirhaleem Feb 20, 2020

amirhaleem Feb 20, 2020

Vagabond Feb 21, 2020

amirhaleem Feb 20, 2020

amirhaleem Feb 20, 2020

lthiery commented Feb 20, 2020

lthiery Feb 21, 2020

Vagabond Feb 21, 2020

evanmcc Feb 22, 2020

Vagabond Feb 22, 2020

evanmcc Feb 22, 2020

Vagabond Feb 22, 2020

evanmcc Feb 22, 2020

Vagabond Feb 22, 2020

PharkMillups commented Jun 12, 2020

jamiew commented Dec 22, 2020


		## What parts of the design do you expect to resolve through the HIP process before this gets merged?

		We need to nail down the permissible sizes of the XOR filters we want to store.


		## What related issues do you consider out of scope for this HIP that could be addressed in the future independently of the solution that comes out of this HIP?

		Pricing and availability of routing table entries for organizations operating


		## What parts of the design do you expect to resolve through the implementation of this feature?

		We need to define where/how these routing entries are stored, how they're

LoRaWAN packet routing (HIP draft) #9

LoRaWAN packet routing (HIP draft) #9

Conversation

Vagabond commented Feb 14, 2020 • edited by amirhaleem Loading

lthiery left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

amirhaleem commented Feb 20, 2020

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

lthiery commented Feb 20, 2020

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

PharkMillups commented Jun 12, 2020

jamiew commented Dec 22, 2020

Vagabond commented Feb 14, 2020 •

edited by amirhaleem

Loading