Skip to main content

Notice: this Wiki will be going read only early in 2024 and edits will no longer be possible. Please see: https://gitlab.eclipse.org/eclipsefdn/helpdesk/-/wikis/Wiki-shutdown-plan for the plan.

Jump to: navigation, search

Difference between revisions of "IoT/LwM2M MQTT Binding/Meeting minutes"

(Actions items)
(Discussion)
Line 16: Line 16:
 
* Marco presented the way Device Management is currently done in Kura:
 
* Marco presented the way Device Management is currently done in Kura:
 
** A request/response model on top of MQTT allows to address each device individually from the cloud, and perform DM commands. Kura allows remote OSGi control with features like start/stop of bundles, remote shell access, ...
 
** A request/response model on top of MQTT allows to address each device individually from the cloud, and perform DM commands. Kura allows remote OSGi control with features like start/stop of bundles, remote shell access, ...
 +
** The request/response model on top of MQTT was presented at EclipseCon: http://youtu.be/EA5j7c2SgqQ
 
** In his spare time Marco has done some experiments with the Leshan Java client to expose OSGi bundles as LwM2M objects. The code is availabe in this Kura branch: https://github.com/eclipse/kura/tree/lwm2m
 
** In his spare time Marco has done some experiments with the Leshan Java client to expose OSGi bundles as LwM2M objects. The code is availabe in this Kura branch: https://github.com/eclipse/kura/tree/lwm2m
 
* Implementing LwM2M over MQTT will be a two-step process:
 
* Implementing LwM2M over MQTT will be a two-step process:
# Implement a request/response model on top of MQTT, providing the ability for an application to send a request to a specific device
+
# Implement a "standard" request/response model on top of MQTT, providing the ability for an application to send a request to a specific device
 
# Implement a REST model (GET, PUT, POST, DELETE) for CRUD operations, including EXECUTE
 
# Implement a REST model (GET, PUT, POST, DELETE) for CRUD operations, including EXECUTE
 
* Peter N. highlights that it would be good to do a gap analysis between the request/response model implemented in MQTT and the CoAP model to make sure all use cases are covered. Similarly, we'll need to check there are no redundancies between the LwM2M and MQTT layers.
 
* Peter N. highlights that it would be good to do a gap analysis between the request/response model implemented in MQTT and the CoAP model to make sure all use cases are covered. Similarly, we'll need to check there are no redundancies between the LwM2M and MQTT layers.
 
* Dave Locke has already been talking with Zach Shelby (ARM) who is involved in the LwM2M specification and he is supportive of a specification for an MQTT binding for LwM2M
 
* Dave Locke has already been talking with Zach Shelby (ARM) who is involved in the LwM2M specification and he is supportive of a specification for an MQTT binding for LwM2M
 +
* Julien reminds that security will be important. Especially in the context of Device Management, we can't afford having other clients eavesdropping MQTT topics...
 
* There were discussions regarding where will be the best place to push the specification of a LwM2M over MQTT binding. The natural place would probably be OMA.   
 
* There were discussions regarding where will be the best place to push the specification of a LwM2M over MQTT binding. The natural place would probably be OMA.   
 
* Marco suggests that we try to have a face-to-face meeting next year.
 
* Marco suggests that we try to have a face-to-face meeting next year.

Revision as of 13:52, 19 December 2014

December 18th conference call

Attendees

  • Shalini Kapur, IBM
  • Julien Vermillard, Sierra Wireless
  • Dave Locke, IBM
  • Peter Crocke, IBM
  • Peter Niblett, IBM
  • Marco Carrer, Eurotech
  • Cristiano de Alti, Eurotech
  • Benjamin Cabé, Eclipse Fdn

Discussion

  • Marco presented the way Device Management is currently done in Kura:
    • A request/response model on top of MQTT allows to address each device individually from the cloud, and perform DM commands. Kura allows remote OSGi control with features like start/stop of bundles, remote shell access, ...
    • The request/response model on top of MQTT was presented at EclipseCon: http://youtu.be/EA5j7c2SgqQ
    • In his spare time Marco has done some experiments with the Leshan Java client to expose OSGi bundles as LwM2M objects. The code is availabe in this Kura branch: https://github.com/eclipse/kura/tree/lwm2m
  • Implementing LwM2M over MQTT will be a two-step process:
  1. Implement a "standard" request/response model on top of MQTT, providing the ability for an application to send a request to a specific device
  2. Implement a REST model (GET, PUT, POST, DELETE) for CRUD operations, including EXECUTE
  • Peter N. highlights that it would be good to do a gap analysis between the request/response model implemented in MQTT and the CoAP model to make sure all use cases are covered. Similarly, we'll need to check there are no redundancies between the LwM2M and MQTT layers.
  • Dave Locke has already been talking with Zach Shelby (ARM) who is involved in the LwM2M specification and he is supportive of a specification for an MQTT binding for LwM2M
  • Julien reminds that security will be important. Especially in the context of Device Management, we can't afford having other clients eavesdropping MQTT topics...
  • There were discussions regarding where will be the best place to push the specification of a LwM2M over MQTT binding. The natural place would probably be OMA.
  • Marco suggests that we try to have a face-to-face meeting next year.

Actions items

  • Marco will send to the group the documentation of the current implementation of DM over MQTT in Kura. This documentation will be publicly available as part of Kura shortly.
  • Marco will initiate a page in the Kura wiki to start collecting the list of DM functions that should initially be implemented in a future PoC
  • Benjamin will organize the next conference call in January.

Back to the top