IETF78 ROLL WG ============== Thanks to Bob Power for taking the notes. 1) WG Status (Chairs - 10 mn) [15] Agenda Bashing --no changes Charter Review & Milestones --------------------------- Good progress since previous WG meeting 2 new RFC's 5826 & 5867 RPL Spec --tremendous progress --a lot of design discipline --achieved goal of minimum viable protocol --produced a solid core spec --AND running code --lot of implementations (approx 15) --Last call issued on July 6, ended on July 27th, major milestone. --track all changes through the ticket mechanism --tool, convenience, does not replace mail lists --helped handle issues in a transparent fashion --full expectation, take clarifications fold them into -11, take this final transition toward RFC stage ROLL Stats --over 10 months, large number of issues opened & closed --currently on 2 remain open Charter Review -------------- Next Steps ---------- LC - tricle ID (July 28 started) - metric ID - OF0 Point-to-point - need to focus on this next Architecture (JP and David will start working on this item) Management model (mibs etc) JP> Re-chartering discussion later: need to take time for gather implementation feed-back before starting new work. Emphasis between here and Beijing is "running code". --a lot of things going forward are optimizations --need a high quality baseline FIRST 2) RPL: Routing Protocol for Low power and Lossy network draft-ietf-roll-rpl-10 RPL Update (Jonathan Hui) ------------------------- Last call opened and closed. 19 tickets opened, all but 2 Major milestones reached, its real! Thanks! Q. when will V11 be posted? A. Today or tomorrow. David> Going forward, we are NOT raising new issues. There may be editorial corrections. Question (?): could you please share some documents, publications on the mechanisms used in RPL on the mailing list? Answer: sure. Security Mechanisms in base spec (Phil Levis) --------------------------------------------- IPR issues - confidentiality, integrity are fine - signature scheme proposed in -10; IPR held by Certicom - terms were not OK - signature scheme is -11 is NOT encumbered (based on RSA) Q. Is there any IPR A. To the best of our knowledge, there is no encumbering IPR ECDSA is more compact; on the NIST recommended list for security schemes --almost surely that we'll come back here --have coding space to deal with more compact one Chairs are not permitted to make IPR claims. --cannot make any judgements --cannot take a legal responsibility --what we can do is observe lack of IPR statements Q. What other kinds of IPRs are related to this draft. A. If there is any IPR that relates to the draft, ANY participat who is aware of IPR, it must be disclosed. Q. Two disclosures were made; 1 from Cisco; 1 related to Certicom Adrian - there is a link on IETF page; that allows you to search; you should search and not ask the chairs. David: Cisco IPR was dealt with by changing of terms. Eliminated the other (Certicom) by removing the IPR 3) The Minimum Rank Objective Function with Hysteresis draft-gnawali-roll-minrank-hysteresis-of-01 (Phil - 10mn) [145] MRHOF (Phil Levis) ------------------ Minimum Rank Objective Function with Hysteresis RPL & Objective Function --ie, what is objective of this instance of rpl (min latency, hop count, etc) Selected Metric - metric to minimize Path Cost - Compute the path metric --at the root -cur_min_path_cost - MIN_PATH_COST -other nodes - min {candidaet_min_path_cost(1..n)} Ex. -etx: paths with smalles expected tx count -latency: least latency cost Hysteresis -metric can have jitter --ETX jitter due to link quality changing --can cause churn -hysteresis delays the effects --short/small changes should not trigger path recomputation -change parent if the new path metric change > certain amount Computing Rank from Cost Metric --energy 255-cost --hop count cost --latency cost/65536 --lq cost --etx cost Rank not definsed for things like throughput; link color; node fanout Pascal: OF could be mixing several metrics together. Could have a basic default way of doing this. Needs to be in OF specification M. Charlser (?) --Q. do we plan to move MRHOF into base draft --A. no Other considerations --Frequency of updates --periodic --reactive --Are metric agnostic parameters possible? --tie breaker for same path cost? M. --if MRHOP is pluggable; it has to be OF=n; MRHOF is a single OF; it will perform A. Wagger Comment: really necessary to define a tie-breaker had a problem in an OLSR implementation JP: yes, but make it optional 4) Routing Metrics used for Path Calculation in Low Power and Lossy Networks (Dominique Ð 10 min) [155] draft-ietf-roll-routing-metrics-08 Dominique Barthel Be sure all metrics have the same sign (increasing is better or worse, but be consistent) PL: rank has fewer bits than latency for example; not scaling them is a problem DB: If we agree on scaling; it will make things simpler If metrics are decoupled, then OF documents should not list type of metrics. --are 5) The Trickle Algorithm - draft-levis-roll-trickle-02 (Phil - 5mn) [160] Quick update on the changes in -02. Currently in LC. JP> ID is a normative reference in RPL core spec (thus we have a downref). Discussion on ID status JP> Will come back to the list to consider status change (informational to standard track) Discussion on complexity Answer: many implementations in the field, takes very limited resources. Thomas CLausen: We have an implementation of RPL. Works fine with trickle just took a few days of coding. 6) Multicast Forwarding Using Trickle - draft-hui-6man-trickle-mcast-00 (TBD - 10mn) [170] Quick overview. JP> Note that the home for this ID is 6man, not ROLL but we found interesting to get an update here. 7 )Reactive Discovery of Point-to-Point Routes in Low Power and Lossy Networks draft-dt-roll-p2p-rpl-01 (Emmanuel - 10mn) [180] Overall discussion. More discussion to take place on the mailing list.