Home
You are not currently signed in.

RFC6206

  1. RFC 6206
Internet Engineering Task Force (IETF)                          P. Levis
Request for Comments: 6206                           Stanford University
Category: Standards Track                                     T. Clausen
ISSN: 2070-1721                                 LIX, Ecole Polytechnique
                                                                  J. Hui
                                                   Arch Rock Corporation
                                                              O. Gnawali
                                                     Stanford University
                                                                   J. Ko
                                                Johns Hopkins University
                                                              March 2011


                         The Trickle Algorithm

Abstract

   The Trickle algorithm allows nodes in a lossy shared medium (e.g.,
   low-power and lossy networks) to exchange information in a highly
   robust, energy efficient, simple, and scalable manner.  Dynamically
   adjusting transmission windows allows Trickle to spread new
   information on the scale of link-layer transmission times while
   sending only a few messages per hour when information does not
   change.  A simple suppression mechanism and transmission point
   selection allow Trickle's communication rate to scale logarithmically
   with density.  This document describes the Trickle algorithm and
   considerations in its use.

Status of This Memo

   This is an Internet Standards Track document.

   This document is a product of the Internet Engineering Task Force
   (IETF).  It represents the consensus of the IETF community.  It has
   received public review and has been approved for publication by the
   Internet Engineering Steering Group (IESG).  Further information on
   Internet Standards is available in Section 2 of RFC 5741.

   Information about the current status of this document, any errata,
   and how to provide feedback on it may be obtained at
   http://www.rfc-editor.org/info/rfc6206.










Levis, et al.                Standards Track                    [Page 1]
RFC 6206                    Trickle Algorithm                 March 2011


Copyright Notice

   Copyright (c) 2011 IETF Trust and the persons identified as the
   document authors.  All rights reserved.

   This document is subject to BCP 78 and the IETF Trust's Legal
   Provisions Relating to IETF Documents
   (http://trustee.ietf.org/license-info) in effect on the date of
   publication of this document.  Please review these documents
   carefully, as they describe your rights and restrictions with respect
   to this document.  Code Components extracted from this document must
   include Simplified BSD License text as described in Section 4.e of
   the Trust Legal Provisions and are provided without warranty as
   described in the Simplified BSD License.

Table of Contents

   1. Introduction ....................................................2
   2. Terminology .....................................................3
   3. Trickle Algorithm Overview ......................................3
   4. Trickle Algorithm ...............................................5
      4.1. Parameters and Variables ...................................5
      4.2. Algorithm Description ......................................5
   5. Using Trickle ...................................................6
   6. Operational Considerations ......................................7
      6.1. Mismatched Redundancy Constants ............................7
      6.2. Mismatched Imin ............................................7
      6.3. Mismatched Imax ............................................8
      6.4. Mismatched Definitions .....................................8
      6.5. Specifying the Constant k ..................................8
      6.6. Relationship between k and Imin ............................8
      6.7. Tweaks and Improvements to Trickle .........................9
      6.8. Uses of Trickle ............................................9
   7. Acknowledgements ...............................................10
   8. Security Considerations ........................................10
   9. References .....................................................11
      9.1. Normative References ......................................11
      9.2. Informative References ....................................11

1.  Introduction

   The Trickle algorithm establishes a density-aware local communication
   primitive with an underlying consistency model that guides when a
   node transmits.  When a node's data does not agree with its
   neighbors, that node communicates quickly to resolve the
   inconsistency (e.g., in milliseconds).  When nodes agree, they slow
   their communication rate exponentially, such that nodes send packets
   very infrequently (e.g., a few packets per hour).  Instead of



Levis, et al.                Standards Track                    [Page 2]
RFC 6206                    Trickle Algorithm                 March 2011


   flooding a network with packets, the algorithm controls the send rate
   so each node hears a small trickle of packets, just enough to stay
   consistent.  Furthermore, by relying only on local communication
   (e.g., broadcast or local multicast), Trickle handles network
   re-population; is robust to network transience, loss, and
   disconnection; is simple to implement; and requires very little
   state.  Current implementations use 4-11 bytes of RAM and are
   50-200 lines of C code [Levis08].

   While Trickle was originally designed for reprogramming protocols
   (where the data is the code of the program being updated), experience
   has shown it to be a powerful mechanism that can be applied to a wide
   range of protocol design problems, including control traffic timing,
   multicast propagation, and route discovery.  This flexibility stems
   from being able to define, on a case-by-case basis, what constitutes
   "agreement" or an "inconsistency"; Section 6.8 presents a few
   examples of how the algorithm can be used.

   This document describes the Trickle algorithm and provides guidelines
   for its use.  It also states requirements for protocol specifications
   that use Trickle.  This document does not provide results regarding
   Trickle's performance or behavior, nor does it explain the
   algorithm's design in detail: interested readers should refer to
   [Levis04] and [Levis08].

2.  Terminology

   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
   "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and
   "OPTIONAL" in this document are to be interpreted as described in
   RFC 2119 [RFC2119].

   Additionally, this document introduces the following terminology:

   Trickle communication rate:  the sum of the number of messages sent
      or received by the Trickle algorithm in an interval.

   Trickle transmission rate:  the sum of the number of messages sent by
      the Trickle algorithm in an interval.

3.  Trickle Algorithm Overview

   Trickle's basic primitive is simple: every so often, a node transmits
   data unless it hears a few other transmissions whose data suggest its
   own transmission is redundant.  Examples of such data include routing
   state, software update versions, and the last heard multicast packet.
   This primitive allows Trickle to scale to thousand-fold variations in
   network density, quickly propagate updates, distribute transmission



Levis, et al.                Standards Track                    [Page 3]
RFC 6206                    Trickle Algorithm                 March 2011


   load evenly, be robust to transient disconnections, handle network
   re-populations, and impose a very low maintenance overhead: one
   example use, routing beacons in the Collection Tree Protocol (CTP)
   [Gnawali09], requires sending on the order of a few packets per hour,
   yet CTP can respond to topology changes in milliseconds.

   Trickle sends all messages to a local communication address.  The
   exact address used can depend on the underlying IP protocol as well
   as how the higher-layer protocol uses Trickle.  In IPv6, for example,
   it can be the link-local multicast address or another local multicast
   address, while in IPv4 it can be the broadcast address
   (255.255.255.255).

   There are two possible results to a Trickle message: either every
   node that hears the message finds that the message data is consistent
   with its own state, or a recipient detects an inconsistency.
   Detection can be the result of either an out-of-date node hearing
   something new, or an updated node hearing something old.  As long as
   every node communicates somehow -- either receives or transmits --
   some node will detect the need for an update.

   For example, consider a simple case where "up to date" is defined by
   version numbers (e.g., network configuration).  If node A transmits
   that it has version V, but B has version V+1, then B knows that A
   needs an update.  Similarly, if B transmits that it has version V+1,
   A knows that it needs an update.  If B broadcasts or multicasts
   updates, then all of its neighbors can receive them without having to
   advertise their need.  Some of these recipients might not have even
   heard A's transmission.  In this example, it does not matter who
   first transmits -- A or B; the inconsistency will be detected in
   either case.

   The fact that Trickle communication can be either transmission or
   reception enables the Trickle algorithm to operate in sparse as well
   as dense networks.  A single, disconnected node must transmit at the
   Trickle communication rate.  In a lossless, single-hop network of
   size n, the Trickle communication rate at each node equals the sum of
   the Trickle transmission rates across all nodes.  The Trickle
   algorithm balances the load in such a scenario, as each node's
   Trickle transmission rate is 1/nth of the Trickle communication rate.
   Sparser networks require more transmissions per node, but the
   utilization of a given broadcast domain (e.g., radio channel over
   space, shared medium) will not increase.  This is an important
   property in wireless networks and other shared media, where the
   channel is a valuable shared resource.  Additionally, reducing
   transmissions in dense networks conserves system energy.





Levis, et al.                Standards Track                    [Page 4]
RFC 6206                    Trickle Algorithm                 March 2011


4.  Trickle Algorithm

   This section describes the Trickle algorithm.

4.1.  Parameters and Variables

   A Trickle timer runs for a defined interval and has three
   configuration parameters: the minimum interval size Imin, the maximum
   interval size Imax, and a redundancy constant k:

   o  The minimum interval size, Imin, is defined in units of time
      (e.g., milliseconds, seconds).  For example, a protocol might
      define the minimum interval as 100 milliseconds.

   o  The maximum interval size, Imax, is described as a number of
      doublings of the minimum interval size (the base-2 log(max/min)).
      For example, a protocol might define Imax as 16.  If the minimum
      interval is 100 ms, then the amount of time specified by Imax is
      100 ms * 65,536, i.e., 6,553.6 seconds or approximately
      109 minutes.

   o  The redundancy constant, k, is a natural number (an integer
      greater than zero).

   In addition to these three parameters, Trickle maintains three
   variables:

   o  I, the current interval size,

   o  t, a time within the current interval, and

   o  c, a counter.

4.2.  Algorithm Description

   The Trickle algorithm has six rules:

   1.  When the algorithm starts execution, it sets I to a value in the
       range of [Imin, Imax] -- that is, greater than or equal to Imin
       and less than or equal to Imax.  The algorithm then begins the
       first interval.

   2.  When an interval begins, Trickle resets c to 0 and sets t to a
       random point in the interval, taken from the range [I/2, I), that
       is, values greater than or equal to I/2 and less than I.  The
       interval ends at I.





Levis, et al.                Standards Track                    [Page 5]
RFC 6206                    Trickle Algorithm                 March 2011


   3.  Whenever Trickle hears a transmission that is "consistent", it
       increments the counter c.

   4.  At time t, Trickle transmits if and only if the counter c is less
       than the redundancy constant k.

   5.  When the interval I expires, Trickle doubles the interval length.
       If this new interval length would be longer than the time
       specified by Imax, Trickle sets the interval length I to be the
       time specified by Imax.

   6.  If Trickle hears a transmission that is "inconsistent" and I is
       greater than Imin, it resets the Trickle timer.  To reset the
       timer, Trickle sets I to Imin and starts a new interval as in
       step 2.  If I is equal to Imin when Trickle hears an
       "inconsistent" transmission, Trickle does nothing.  Trickle can
       also reset its timer in response to external "events".

   The terms "consistent", "inconsistent", and "events" are in quotes
   because their meaning depends on how a protocol uses Trickle.

   The only time the Trickle algorithm transmits is at step 4 of the
   above algorithm.  This means there is an inherent delay between
   detecting an inconsistency (shrinking I to Imin) and responding to
   that inconsistency (transmitting at time t in the new interval).
   This is intentional.  Immediately responding to detecting an
   inconsistency can cause a broadcast storm, where many nodes respond
   at once and in a synchronized fashion.  By making responses follow
   the Trickle algorithm (with the minimal interval size), a protocol
   can benefit from Trickle's suppression mechanism and scale across a
   huge range of node densities.

5.  Using Trickle

   A protocol specification that uses Trickle MUST specify:

   o  Default values for Imin, Imax, and k.  Because link layers can
      vary widely in their properties, the default value of Imin SHOULD
      be specified in terms of the worst-case latency of a link-layer
      transmission.  For example, a specification should say "the
      default value of Imin is 4 times the worst-case link-layer
      latency" and should not say "the default value of Imin is
      500 milliseconds".  Worst-case latency is approximately the time
      until the first link-layer transmission of the frame, assuming an
      idle channel (does not include backoff, virtual carrier sense,
      etc.).

   o  What constitutes a "consistent" transmission.



Levis, et al.                Standards Track                    [Page 6]
RFC 6206                    Trickle Algorithm                 March 2011


   o  What constitutes an "inconsistent" transmission.

   o  What "events", if any -- besides inconsistent transmissions --
      reset the Trickle timer.

   o  What information a node transmits in Trickle messages.

   o  What actions outside the algorithm the protocol takes, if any,
      when it detects an inconsistency.

6.  Operational Considerations

   It is RECOMMENDED that a protocol that uses Trickle include
   mechanisms to inform nodes of configuration parameters at runtime.
   However, it is not always possible to do so.  In the cases where
   different nodes have different configuration parameters, Trickle may
   have unintended behaviors.  This section outlines some of those
   behaviors and operational considerations as educational exercises.

6.1.  Mismatched Redundancy Constants

   If nodes do not agree on the redundancy constant k, then nodes with
   higher values of k will transmit more often than nodes with lower
   values of k.  In some cases, this increased load can be independent
   of the density.  For example, consider a network where all nodes but
   one have k=1, and this one node has k=2.  The different node can end
   up transmitting on every interval: it is maintaining a Trickle
   communication rate of 2 with only itself.  Hence, the danger of
   mismatched k values is uneven transmission load that can deplete the
   energy of some nodes in a low-power network.

6.2.  Mismatched Imin

   If nodes do not agree on Imin, then some nodes, on hearing
   inconsistent messages, will transmit sooner than others.  These
   faster nodes will have their intervals grow to a size similar to that
   of the slower nodes within a single slow interval time, but in that
   period may suppress the slower nodes.  However, such suppression will
   end after the first slow interval, when the nodes generally agree on
   the interval size.  Hence, mismatched Imin values are usually not a
   significant concern.  Note that mismatched Imin values and matching
   Imax doubling constants will lead to mismatched maximum interval
   lengths.








Levis, et al.                Standards Track                    [Page 7]
RFC 6206                    Trickle Algorithm                 March 2011


6.3.  Mismatched Imax

   If nodes do not agree on Imax, then this can cause long-term problems
   with transmission load.  Nodes with small Imax values will transmit
   faster, suppressing those with larger Imax values.  The nodes with
   larger Imax values, always suppressed, will never transmit.  In the
   base case, when the network is consistent, this can cause long-term
   inequities in energy cost.

6.4.  Mismatched Definitions

   If nodes do not agree on what constitutes a consistent or
   inconsistent transmission, then Trickle may fail to operate properly.
   For example, if a receiver thinks a transmission is consistent, but
   the transmitter (if in the receiver's situation) would have thought
   it inconsistent, then the receiver will not respond properly and
   inform the transmitter.  This can lead the network to not reach a
   consistent state.  For this reason, unlike the configuration
   constants k, Imin, and Imax, consistency definitions MUST be clearly
   stated in the protocol and SHOULD NOT be configured at runtime.

6.5.  Specifying the Constant k

   There are some edge cases where a protocol may wish to use Trickle
   with its suppression disabled (k is set to infinity).  In general,
   this approach is highly dangerous and it is NOT RECOMMENDED.
   Disabling suppression means that every node will always send on every
   interval; this can lead to congestion in dense networks.  This
   approach is especially dangerous if many nodes reset their intervals
   at the same time.  In general, it is much more desirable to set k to
   a high value (e.g., 5 or 10) than infinity.  Typical values for k
   are 1-5: these achieve a good balance between redundancy and low cost
   [Levis08].

   Nevertheless, there are situations where a protocol may wish to turn
   off Trickle suppression.  Because k is a natural number
   (Section 4.1), k=0 has no useful meaning.  If a protocol allows k to
   be dynamically configured, a value of 0 remains unused.  For ease of
   debugging and packet inspection, having the parameter describe k-1
   rather than k can be confusing.  Instead, it is RECOMMENDED that
   protocols that require turning off suppression reserve k=0 to mean
   k=infinity.

6.6.  Relationship between k and Imin

   Finally, a protocol SHOULD set k and Imin such that Imin is at least
   two to three times as long as it takes to transmit k packets.
   Otherwise, if more than k nodes reset their intervals to Imin, the



Levis, et al.                Standards Track                    [Page 8]
RFC 6206                    Trickle Algorithm                 March 2011


   resulting communication will lead to congestion and significant
   packet loss.  Experimental results have shown that packet losses from
   congestion reduce Trickle's efficiency [Levis04].

6.7.  Tweaks and Improvements to Trickle

   Trickle is based on a small number of simple, tightly integrated
   mechanisms that are highly robust to challenging network
   environments.  In our experiences using Trickle, attempts to tweak
   its behavior are typically not worth the cost.  As written, the
   algorithm is already highly efficient: further reductions in
   transmissions or response time come at the cost of failures in edge
   cases.  Based on our experiences, we urge protocol designers to
   suppress the instinct to tweak or improve Trickle without a great
   deal of experimental evidence that the change does not violate its
   assumptions and break the algorithm in edge cases.

   With this warning in mind, Trickle is far from perfect.  For example,
   Trickle suppression typically leads sparser nodes to transmit more
   than denser ones; it is far from the optimal computation of a minimum
   cover.  However, in dynamic network environments such as wireless and
   low-power, lossy networks, the coordination needed to compute the
   optimal set of transmissions is typically much greater than the
   benefits it provides.  One of the benefits of Trickle is that it is
   so simple to implement and requires so little state yet operates so
   efficiently.  Efforts to improve it should be weighed against the
   cost of increased complexity.

6.8.  Uses of Trickle

   The Trickle algorithm has been used in a variety of protocols, in
   operational as well as academic settings.  Giving a brief overview of
   some of these uses provides useful examples of how and when it can be
   used.  These examples should not be considered exhaustive.

   Reliable flooding/dissemination: A protocol uses Trickle to
   periodically advertise the most recent data it has received,
   typically through a version number.  An inconsistency occurs when a
   node hears a newer version number or receives new data.  A
   consistency occurs when a node hears an older or equal version
   number.  When hearing an older version number, rather than reset its
   own Trickle timer, the node sends an update.  Nodes with old version
   numbers that receive the update will then reset their own timers,
   leading to fast propagation of the new data.  Examples of this use
   include multicast [Hui08a], network configuration [Lin08] [Dang09],
   and installing new application programs [Hui04] [Levis04].





Levis, et al.                Standards Track                    [Page 9]
RFC 6206                    Trickle Algorithm                 March 2011


   Routing control traffic: A protocol uses Trickle to control when it
   sends beacons that contain routing state.  An inconsistency occurs
   when the routing topology changes in a way that could lead to loops
   or significant stretch: examples include when the routing layer
   detects a routing loop or when a node's routing cost changes
   significantly.  Consistency occurs when the routing topology is
   operating well and is delivering packets successfully.  Using the
   Trickle algorithm in this way allows a routing protocol to react very
   quickly to problems (Imin is small) but send very few beacons when
   the topology is stable.  Examples of this use include the IPv6
   routing protocol for low-power and lossy networks (RPL) [RPL], CTP
   [Gnawali09], and some current commercial IPv6 routing layers
   [Hui08b].

7.  Acknowledgements

   The authors would like to acknowledge the guidance and input provided
   by the ROLL chairs, David Culler and JP Vasseur.

   The authors would also like to acknowledge the helpful comments of
   Yoav Ben-Yehezkel, Alexandru Petrescu, and Ulrich Herberg, which
   greatly improved the document.

8.  Security Considerations

   As it is an algorithm, Trickle itself does not have any specific
   security considerations.  However, two security concerns can arise
   when Trickle is used in a protocol.  The first is that an adversary
   can force nodes to send many more packets than needed by forcing
   Trickle timer resets.  In low-power networks, this increase in
   traffic can harm system lifetime.  The second concern is that an
   adversary can prevent nodes from reaching consistency.

   Protocols can prevent adversarial Trickle resets by carefully
   selecting what can cause a reset and protecting these events and
   messages with proper security mechanisms.  For example, if a node can
   reset nearby Trickle timers by sending a certain packet, this packet
   should be authenticated such that an adversary cannot forge one.

   An adversary can possibly prevent nodes from reaching consistency by
   suppressing transmissions with "consistent" messages.  For example,
   imagine node A detects an inconsistency and resets its Trickle timer.
   If an adversary can prevent A from sending messages that inform
   nearby nodes of the inconsistency in order to repair it, then A may
   remain inconsistent indefinitely.  Depending on the security model of
   the network, authenticated messages or a transitive notion of
   consistency can prevent this problem.  For example, let us suppose an
   adversary wishes to suppress A from notifying neighbors of an



Levis, et al.                Standards Track                   [Page 10]
RFC 6206                    Trickle Algorithm                 March 2011


   inconsistency.  To do so, it must send messages that are consistent
   with A.  These messages are by definition inconsistent with those of
   A's neighbors.  Correspondingly, an adversary cannot simultaneously
   prevent A from notifying neighbors and not notify the neighbors
   itself (recall that Trickle operates on shared, broadcast media).
   Note that this means Trickle should filter unicast messages.

9.  References

9.1.  Normative References

   [RFC2119]  Bradner, S., "Key words for use in RFCs to Indicate
              Requirement Levels", BCP 14, RFC 2119, March 1997.

9.2.  Informative References

   [Dang09]   Dang, T., Bulusu, N., Feng, W., and S. Park, "DHV: A Code
              Consistency Maintenance Protocol for Multi-hop Wireless
              Networks", Wireless Sensor Networks: 6th European
              Conference Proceedings EWSN 2009 Cork, February 2009,
              <http://portal.acm.org/citation.cfm?id=1506781>.

   [Gnawali09]
              Gnawali, O., Fonseca, R., Jamieson, K., Moss, D., and P.
              Levis, "Collection Tree Protocol", Proceedings of the 7th
              ACM Conference on Embedded Networked Sensor
              Systems, SenSys 2009, November 2009,
              <http://portal.acm.org/citation.cfm?id=1644038.1644040>.

   [Hui04]    Hui, J. and D. Culler, "The dynamic behavior of a data
              dissemination protocol for network programming at scale",
              Proceedings of the 2nd ACM Conference on Embedded
              Networked Sensor Systems, SenSys 2004, November 2004,
              <http://portal.acm.org/citation.cfm?id=1031506>.

   [Hui08a]   Hui, J., "An Extended Internet Architecture for Low-Power
              Wireless Networks - Design and Implementation", UC
              Berkeley Technical Report EECS-2008-116, September 2008,
              <http://www.eecs.berkeley.edu/Pubs/>.

   [Hui08b]   Hui, J. and D. Culler, "IP is dead, long live IP for
              wireless sensor networks", Proceedings of the 6th ACM
              Conference on Embedded Networked Sensor Systems, SenSys
              2008, November 2008,
              <http://portal.acm.org/citation.cfm?id=1460412.1460415>.






Levis, et al.                Standards Track                   [Page 11]
RFC 6206                    Trickle Algorithm                 March 2011


   [Levis04]  Levis, P., Patel, N., Culler, D., and S. Shenker,
              "Trickle: A Self-Regulating Algorithm for Code Propagation
              and Maintenance in Wireless Sensor Networks", Proceedings
              of the First USENIX/ACM Symposium on Networked Systems
              Design and Implementation, NSDI 2004, March 2004,
              <http://portal.acm.org/citation.cfm?id=1251177>.

   [Levis08]  Levis, P., Brewer, E., Culler, D., Gay, D., Madden, S.,
              Patel, N., Polastre, J., Shenker, S., Szewczyk, R., and A.
              Woo, "The Emergence of a Networking Primitive in Wireless
              Sensor Networks", Communications of the ACM, Vol. 51 No.
              7, July 2008,
              <http://portal.acm.org/citation.cfm?id=1364804>.

   [Lin08]    Lin, K. and P. Levis, "Data Discovery and Dissemination
              with DIP", Proceedings of the 7th international conference
              on Information processing in sensor networks, IPSN 2008,
              April 2008,
              <http://portal.acm.org/citation.cfm?id=1371607.1372753>.

   [RPL]      Winter, T., Ed., Thubert, P., Ed., Brandt, A., Clausen,
              T., Hui, J., Kelsey, R., Levis, P., Pister, K., Struik,
              R., and JP. Vasseur, "RPL: IPv6 Routing Protocol for Low
              power and Lossy Networks", Work in Progress, March 2011.



























Levis, et al.                Standards Track                   [Page 12]
RFC 6206                    Trickle Algorithm                 March 2011


Authors' Addresses

   Philip Levis
   Stanford University
   358 Gates Hall
   Stanford, CA  94305
   USA

   Phone: +1 650 725 9064
   EMail: pal@cs.stanford.edu


   Thomas Heide Clausen
   LIX, Ecole Polytechnique

   Phone: +33 6 6058 9349
   EMail: T.Clausen@computer.org


   Jonathan Hui
   Arch Rock Corporation
   501 2nd St., Suite 410
   San Francisco, CA  94107
   USA

   EMail: jhui@archrock.com


   Omprakash Gnawali
   Stanford University
   S255 Clark Center, 318 Campus Drive
   Stanford, CA  94305
   USA

   Phone: +1 650 725 6086
   EMail: gnawali@cs.stanford.edu


   JeongGil Ko
   Johns Hopkins University
   3400 N. Charles St., 224 New Engineering Building
   Baltimore, MD  21218
   USA

   Phone: +1 410 516 4312
   EMail: jgko@cs.jhu.edu





Levis, et al.                Standards Track                   [Page 13]
  1. RFC 6206