Inter-network Coexistence in the Internet of Things
draft-feeney-t2trg-inter-network-00

Network Working Group                                          L. Feeney
Internet-Draft                                        Uppsala University
Intended status: Informational                                  V. Fodor
Expires: January 4, 2018                                             KTH
                                                           July 03, 2017

          Inter-network Coexistence in the Internet of Things
                  draft-feeney-t2trg-inter-network-00

Abstract

   The breadth of IoT applications implies that future wireless
   environments will be characterized by the presence of many diverse,
   administratively independent IoT networks operating in the same
   physical location.  In many cases, these networks will use unlicensed
   spectrum, due to its low cost and ease of deployment.  However, this
   spectrum is becoming increasingly crowded.  IoT networks will
   therefore be subject to wireless interference, both from similar
   networks and from networks that use the channel in very different
   ways.

   To date, there have been few studies or testbeds that fully reflect
   this aspect of the future IoT operating environment.  This document
   describes some of the main issues in network co-existence in IoT
   environments, focusing on protocol-level interactions.  It identifies
   two issues for the IRTF t2trg community.  The first is to define best
   practices for performance evaluation and protocol design in the
   context of inter-network interference.  The second is the potential
   use of higher layer protocols to actively participate in interference
   mitigation.

Status of This Memo

   This Internet-Draft is submitted in full conformance with the
   provisions of BCP 78 and BCP 79.

   Internet-Drafts are working documents of the Internet Engineering
   Task Force (IETF).  Note that other groups may also distribute
   working documents as Internet-Drafts.  The list of current Internet-
   Drafts is at http://datatracker.ietf.org/drafts/current/.

   Internet-Drafts are draft documents valid for a maximum of six months
   and may be updated, replaced, or obsoleted by other documents at any
   time.  It is inappropriate to use Internet-Drafts as reference
   material or to cite them other than as "work in progress."

   This Internet-Draft will expire on January 4, 2018.

Feeney & Fodor           Expires January 4, 2018                [Page 1]
Internet-Draft          inter-network coexistence              July 2017

Copyright Notice

   Copyright (c) 2017 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.  IoT interference challenges . . . . . . . . . . . . . . . . .   3
     2.1.  Scale . . . . . . . . . . . . . . . . . . . . . . . . . .   3
     2.2.  Independence  . . . . . . . . . . . . . . . . . . . . . .   3
     2.3.  Battery lifetime  . . . . . . . . . . . . . . . . . . . .   4
     2.4.  Resource constraints  . . . . . . . . . . . . . . . . . .   4
     2.5.  Diversity . . . . . . . . . . . . . . . . . . . . . . . .   4
   3.  Interaction behaviors . . . . . . . . . . . . . . . . . . . .   5
   4.  Network co-existence in the IRTF/IETF context . . . . . . . .   6
     4.1.  Responding to link layer evolution  . . . . . . . . . . .   7
     4.2.  Protocol evaluation . . . . . . . . . . . . . . . . . . .   7
     4.3.  Active mitigation strategies  . . . . . . . . . . . . . .   7
   5.  Security Considerations . . . . . . . . . . . . . . . . . . .   9
   6.  Conclusion  . . . . . . . . . . . . . . . . . . . . . . . . .  10
   7.  Informative References  . . . . . . . . . . . . . . . . . . .  10
   Acknowledgements  . . . . . . . . . . . . . . . . . . . . . . . .  11
   Authors' Addresses  . . . . . . . . . . . . . . . . . . . . . . .  11

1.  Introduction

   An enormous range of IoT applications are expected to become
   pervasive in daily life.  Networks will be installed in public
   spaces, businesses, and residences by a wide range of individual,
   commercial, and government actors.  This means that there will be
   many diverse, administratively independent networks operating in the
   same physical location.  For example, a future home environment may
   include IoT applications for security, heating and cooling, elder
   care, air quality monitoring, personal health and fitness, smart home
   appliances, structural monitoring, lighting, utilities, and
   entertainment.

Feeney & Fodor           Expires January 4, 2018                [Page 2]
Internet-Draft          inter-network coexistence              July 2017

   Many of these applications will use unlicensed spectrum due to low
   cost and simplicity of deployment for both the user and developer.
   In unlicensed spectrum, there is no authority that has a management
   relationship with (or even knows about) all of the potentially
   interfering networks that can be present in some location.  This
   means that there is no entity that can coordinate networks' use of
   the shared wireless channel.  Networks will therefore experience
   interference caused by transmissions from devices belonging to other
   networks.

   Much existing work in inter-network interference is analytic, based
   on statistical distributions of interfering waveforms under various
   conditions (see e.g.  [NIST]).  There have also been a few papers
   proposing interference mitigation strategies for interference between
   specific protocols (see e.g.  [SURVEY]), particularly between IEEE
   802.11 and IEEE 802.15.4.  In general, existing standards and
   products tend to rely on frequency agility and low duty cycles to
   avoid interference.  To date, however, there have been very few
   studies or testbeds that fully reflect the complex interference
   scenarios of the future IoT environment, particularly in the context
   of protocol-level interactions.

2.  IoT interference challenges

   The widespread deployment and diversity of IoT networks will create
   new challenges in managing network coexistence.

2.1.  Scale

   As IoT becomes pervasive, there may be a large number of networks and
   devices operating in any given location.  Devices from the various
   networks will be topologically inter-mingled.  Interaction scenarios
   will also be highly dynamic, with mobility leading to frequent
   changes in the set of interfering devices.

2.2.  Independence

   In unlicensed spectrum, there is not necessarily any trust
   relationship between networks.  Networks with overlapping
   transmission footprints may well have been deployed by different
   actors (e.g. in adjacent apartments).  There is no single authority
   that has an administrative relationship with all of the potentially
   interfering networks in some location.  Devices within a network will
   be able to authenticate themselves to each other, but the network
   itself may not have any meaningful external identity.

   This means that there is no entity that all networks can trust to
   coordinate their access to the shared channel.

Feeney & Fodor           Expires January 4, 2018                [Page 3]
Internet-Draft          inter-network coexistence              July 2017

2.3.  Battery lifetime

   For battery-powered devices, maximizing lifetime is essential.
   System design is often dominated by the need to minimize device
   activity and especially by the need to keep the radio turned off as
   much as possible.

   This means that there must be some way for senders and receivers to
   coordinate their radio operations.  These mechanisms often depend
   heavily on careful timing of radio operations, in addition to (or
   instead of) exchanging control packets.  Examples include the
   synchronized beacon-enabled PAN, asynchronous wakeup based on low
   power listening (e.g.  ContikiMAC), and TSCH-based scheduled mesh
   (e.g.  WirelessHART).  This timing dependence means that may be more
   sensitive to disruption than might be expected from just considering
   overall channel utilization and collision probabilities [FF15].

   Battery constraints also severely limit devices' ability to listen to
   the channel to observe the behavior of potentially interfering
   networks.

2.4.  Resource constraints

   Iot networks may be severely resource constrained.  Channel capacity
   and battery limitations have been discussed above.  It is also common
   that IoT devices have very limited on-board CPU and memory.  For many
   IoT applications, devices must be very low-cost and easily deployed
   and managed by non-expert users.  These factors severely constrain
   the design space and limit the complexity of any protocol for this
   environment.

2.5.  Diversity

   Although even identical networks experience inter-network
   interference, diversity of radios, protocols, and applications
   creates additional challenges.  This diversity is fundamentally due
   to the diversity of the IoT application requirements, and therefore
   convergence to a single solution is unlikely.

   Different kinds of radios use different modulation schemes to encode
   data on the channel, resulting different patterns of radio energy
   distributed over time and spectrum.  They also divide the spectrum
   into channels differently.  This means that network devices may not
   be able to directly identify the kind of radio that is causing
   interference and packet loss.  In some cases, it may not be possible
   for channel sensing mechanisms to reliably detect the presence of
   interfering transmissions.

Feeney & Fodor           Expires January 4, 2018                [Page 4]
Internet-Draft          inter-network coexistence              July 2017

   Different radio technologies are intended to provide different data
   rates, transmission powers, and coverage areas.  In 2.4GHz spectrum,
   IEEE 802.15.4 is a low power, low data-rate radio; its short (128B)
   frames, transmitted at 250kbps, occupy the channel for up to a few
   ms..  IEEE 802.11 networks have much higher data rates (10's of Mbps)
   and (usually) higher transmit powers.  In sub-GHz spectrum, IEEE
   802.15.4g/Wi-SUN (smart utility network) radios has data rates 50-200
   kbps to communicate over distances of up to ~1km, while LoRa radios
   use data rates of <5 kbps to provide communication over 10's of km.

   These differences imply large differences in packet transmission
   times, which can range from <1 ms to 100's of ms.  Timing parameters
   in a given MAC layer, such as backoff and retransmission intervals,
   tend to be proportional to packet transmission times for the
   underlying PHY.  This can reduce the effectiveness of backoff
   strategies in mixed radio environments, though it can also provide
   opportunities for co-existence strategies (e.g. ).

   Even where networks use the same radio and PHY they may only share
   part of their protocol stack.  The range of protocols based on the
   IEEE 802.15.4 PHY layer is a case in point.  It includes beacon-
   enabled star-topologies (with both CSMA and TDMA/GTS modes),
   scheduled mesh (e.g.  IEEE 802.15.4 TSCH, WirelessHART and ISA
   100.11a standards) and decentralized, asynchronous low power MACs
   (e.g.  ContikiMAC).

   Although IP(v6) is widely (though not universally) used in IEEE
   802.15.4-based network stacks, there is also considerable diversity
   in higher layer protocols.  Some networks are unrouted star-
   topologies, while others use multi-hop routing protocols, such as RPL
   or SmartMesh.  There is also diversity in transport protocols (e.g.
   TCP, CoAP) and applications (many highly specialized).

   IoT networks therefore have a wide range of channel utilization
   patterns.  These include synchronous and asynchronous wakeup,
   periodic announcements and data collection, multi-hop forwarding, and
   bursty responses to detected events.  Communication requirements also
   vary: While few IoT applications are intended as real time control
   systems, many are based on "reasonably" reliable and timely delivery
   of small amounts of sensor data and control traffic.

3.  Interaction behaviors

   Interference between WiFi networks is widely observed, especially in
   dense residential and urban areas, where there are many independently
   deployed networks.  Wifi is an example of a strongly homogeneous
   interference environment.  Most WiFi networks consist of an AP and
   associated devices that communicate directly with their AP.  This

Feeney & Fodor           Expires January 4, 2018                [Page 5]
Internet-Draft          inter-network coexistence              July 2017

   simplifies intra-network coordination and tends to create a somewhat
   cellular topology.  There is also a strong convention for
   interference-reducing channel assignments (channels 1, 6, and 11 in
   2.4GHz spectrum).  Nearly all WiFi devices use the same CSMA-based
   MAC and traffic is dominated by one application - media streaming -
   which is supported by adaptive mechanisms throughout the protocol
   stack.  Despite these simplifying factors, interference is still
   considered a problem, even by the general public.

   2.4GHz spectrum is also a heterogeneous interference environment.  A
   common scenario will involve high-power, high-traffic WiFi networks
   impacting networks based on low-power, low-bitrate radios, such as
   IEEE 802.15.4.  This case has been widely studied and several
   specific protocol variations have been proposed (e.g.  [JP14]), but
   practical solutions mostly involve identifying and using least
   interfered channels.  But in places where there is a lot of WiFi
   traffic, there may be very few such channels.  As a result, the
   various low-power IoT networks operating in these areas may be
   crowded into a very small number of "good" channels.

   This leads to a variety of interactions among IEEE 802.15.4 based
   protocols.  For example, two networks may allocate TDMA transmission
   slots that conflict with each other.  This is because it not possible
   for independent networks to explicitly coordinate their slot
   allocations, while battery constraints make it hard to do a lot of
   channel sensing.

   Recent results [FF15] show that interference between both TDMA and
   CSMA-based IEEE 802.15.4 beacon-enabled PANs can lead to
   synchronization/ desynchronization patterns and episodes of severe
   packet loss - even when the channel itself is only lightly loaded.
   These outages can, in turn, affect the operation of higher-layer
   protocols.  More generally, this shows the negative impact of
   periodic behavior, which is intended to reduce the radio duty cycle
   and hence energy consumption.

4.  Network co-existence in the IRTF/IETF context

   A recent, very broad survey of spectrum sharing research is found in
   [SURVEY].  To date, there have been few studies of the interference
   scenarios outlined above, particularly at the scale and diversity
   that are expected in future IoT scenarios.  In addition, there are
   very few testbeds or simulation tools that are intended to reflect
   the future IoT operating environment.

   A discussion of the challenges of testing wireless co-existence for
   both licensed and unlicensed spectrum is found in NIST.  However,
   this document only considers only radio PHY layer interactions.

Feeney & Fodor           Expires January 4, 2018                [Page 6]
Internet-Draft          inter-network coexistence              July 2017

   There is a substantial need for better understanding of network co-
   existence among the most widely used IoT technologies.  This is an
   open research problem for practical deployment of IoT networks, with
   several areas of relevance for IETF/IRTF activities.  In particular,
   we highlight 1) developing best practices for protocol performance
   evaluation and 2) research into higher layer protocols for explicit
   network coordination as issues that are particularly relevant to IRTF
   t2trg.

4.1.  Responding to link layer evolution

   Radio technologies and link layer standards will continue to evolve
   to provide increased resilience to interference.  Such developments
   are naturally relevant to IETF activities regarding the definition of
   IPv6 over various IoT link layers (e.g. 6low, 6tisch) and closely
   related higher layers protocols such as RPL and CoAP.

4.2.  Protocol evaluation

   Maximizing channel and battery efficiency and minimizing the impact
   of both intra- and inter-network interference is largely in the
   domain of spectrum regulation and the PHY/MAC layer and therefore
   out-of-scope for IETF/IRTF.

   Nevertheless, higher-layer protocols are also affected by - and can
   contribute to - adverse interactions between networks sharing the
   wireless channel.  For example, they may have adaptive behaviors or
   timing dependencies that are sensitive to patterns of loss and delay
   created by inter-network interference.  Or they may have periodic or
   bursty communication patterns that contribute to adverse
   interactions.  The performance of existing (and forthcoming) IETF
   protocols such as 6LoWPAN, 6tisch, 6lpwan, RPL, and CoAP under
   complex IoT interference scenarios is not well understood.

   We therefore argue that performance evaluation of IoT protocols
   should consider whether they will perform acceptably in the presence
   of diverse networks operating in the same spectrum.  This is a poorly
   understood area and there is a lack of simulation or testbed
   environments that provide scale and diversity characteristic of
   future IoT environments.  Documentation and advocacy of best
   practices for protocol evaluation scenarios is therefore relevant to
   IRTF t2trg activities.

4.3.  Active mitigation strategies

   Interference mitigation is likely to rely primarily on improved
   resilience and local adaptation in PHY and MAC layer protocols and
   (to a lesser extent) in higher layer protocols.  However, there may

Feeney & Fodor           Expires January 4, 2018                [Page 7]
Internet-Draft          inter-network coexistence              July 2017

   be opportunities for high layer protocols to actively participate in
   interference mitigation by enabling explicit coordination between
   networks.

   When two networks use the same radio and channel, it may be possible
   for frames transmitted by devices in one network to be successfully
   received by devices in other networks.  These external frames cannot
   be authenticated and are generally discarded at the MAC layer.
   However, they might provide a way for networks to exchange signaling
   information at the network layer, despite their diverse channel
   access or higher-layer protocols.  This could be used for devices to
   announce their expected channel utilization, for example.

   There are substantial challenges in developing such a mechanism:

   1) There is an enormous diversity of radios, channel access methods
   and utilization patterns that might need to be described.  It is not
   clear what information should be signaled or what actions a receiver
   should take in response.

   2) Battery lifetime, channel capacity, and device CPU / memory
   resources continue to be significant limitations.  In particular, the
   radio duty cycle is highly constrained, limiting both sensing and
   communication..

   3) Any such mechanism must operate effectively in the absence of any
   administrative or trust relationship between networks.  Any proposed
   solution will therefore need to be resilient to the possibility of
   incompatible, oblivious, selfish, or even hostile networks
   participating (or not) in some mitigation mechanism.  (See Security
   Considerations.)

   4) The privacy implications of networks sharing information about
   their activity must be carefully considered.  (See Security
   Considerations.)

   This remains a very open research area and one that we argue is
   particularly amenable to standards and interoperability oriented
   approaches enabled by IRTF t2trg.  There may be synergy with IRTF
   t2trg work in IoT semantic interoperability, allowing IoT networks to
   describe not only the 'things' they connect, but also themselves.
   There may also be synergy with IETF activities (e.g. spud/plus) in
   making signaling information available within encrypted flows.

   It is also possible to consider the possibility of coordination
   between networks that use different radios and cannot exchange
   packets Specialized techniques for very simple low bitrate signaling
   between networks using different radios have been proposed.  It is

Feeney & Fodor           Expires January 4, 2018                [Page 8]
Internet-Draft          inter-network coexistence              July 2017

   also possible to consider explicit mitigation strategies that enable
   information exchange via some service coordinated in the Internet
   infrastructure.  Inspiration here is from cognitive radio solutions
   where secondary users obtain information about activity of primary
   users from trusted sources.

5.  Security Considerations

   This document focuses on co-existence between independently
   administrated networks operating in the same location.  The biggest
   security challenge is therefore that such networks do not necessarily
   have any basis for a trust relationship.

   Regulations concerning unlicensed spectrum control radio behaviors
   such as power spectral density, channelization, or duty cycle.  They
   do not mandate the use of any specific protocol, nor is it possible
   to ensure that a potentially interfering network is correctly
   implementing any particular co-existence mechanism.

   Any proposed solution will therefore need to be resilient to the
   possibility of incompatible, oblivious, selfish, or even hostile
   networks participating (or not) in some interference mitigation
   mechanism.  This is especially true for methods in which networks
   actively coordinate their use of the shared channel.  At a minimum,
   participating in information exchange should not substantially
   increase vulnerability to disruption in the case of a malicious (or
   merely incompatible) actor.

   IoT networks that try to be friendly toward each other may disclose
   substantial information about their operation.  There are privacy
   issues associated with IoT networks making such information visible,
   because of their close coupling with human activity.  Particularly
   for health-related applications, even being able to identify the type
   of network application or its level of activity may reveal sensitive
   information.  Ideally, it should be possible for a network to both
   obfuscate its communication patterns (if needed) and to cooperate in
   minimizing adverse interactions.

   One maxim that may be useful in designing the set of information that
   a network discloses as a matter of course with the intention of
   facilitating coexistence is that the information disclosed should not
   provide more insight than that information an attacker might have
   gained by simply observing the network for a while.  But note that
   simply disclosing that information in an accessible way still changes
   the economy of surveillance -- the objective is that it also changes
   the economy of coexistence, and these effects need to be carefully
   weighed against each other.

Feeney & Fodor           Expires January 4, 2018                [Page 9]
Internet-Draft          inter-network coexistence              July 2017

6.  Conclusion

   Understanding and mitigating the impact of inter-network interference
   on performance and reliability is essential for successful large-
   scale deployment of IoT solutions.  Particularly in residential and
   urban environments, high density of WiFi networks will limit the
   number of "good" channels available to low-power IoT networks.  As a
   result, administratively independent IoT networks - possibly with
   quite different channel access behavior - are likely to operate on
   shared channels.

   Potential interactions among these networks are not well understood.
   For example: What happens if two or more independent networks using
   CoAP over RPL over 6LowPAN (or 6tisch) over IEEE 1802.15.4 are
   operating in the same room?  What happens if a beacon-enabled PAN (or
   a Thread or ZigBee or ContikiMAC network, etc.) is thrown into the
   mix?  Especially in a WiFi heavy environment, the value of channel
   hopping for interference mitigation in IEEE 802.15.4 networks may be
   limited.  Similarly, how will LPWAN networks such as LoRa and SigFox,
   with coverage areas of 10's of km sq., interact with other sub-GHz
   networks such as Wi-SUN/IEEE 802.15.4g?

   Interference mitigation is largely the domain of spectrum regulation
   and the PHY/MAC layers.  There are IETF/IRTF interests as well, most
   obviously for IoT protocols such as 6LowPAN, 6tisch, RPL, and CoAP.
   Two open issues are especially relevant to IRTF t2trg: Many
   interference scenarios are not well understood, particularly with
   regard to protocol-level interactions.  Best practices for
   performance evaluation should be developed to reflect future IoT
   environments.  There may also be opportunities for active
   interference mitigation via explicit coordination and information
   sharing, topics which are particularly amenable to interoperability
   and standards oriented research.  However, there are substantial
   research challenges.

7.  Informative References

   [FF15]     Feeney, L. and V. Fodor, "Reliability in co-located
              802.15.4 personal area networks", Proceedings of the 6th
              ACM International Workshop on Pervasive Wireless
              Healthcare - MobiHealth '16 , DOI 10.1145/2944921.2944923,
              2016.

   [IEEE802154]
              "IEEE Standard for Low-Rate Wireless Networks",
              IEEE standard, DOI 10.1109/ieeestd.2016.7460875, n.d..

Feeney & Fodor           Expires January 4, 2018               [Page 10]
Internet-Draft          inter-network coexistence              July 2017

   [JP14]     Javed, Q. and R. Prakash, "CHAMELEON: A Framework for
              Coexistence of Wireless Technologies in an Unlicensed
              Band", Wireless Personal Communications Vol. 77, pp.
              777-808, DOI 10.1007/s11277-013-1536-7, November 2013.

   [NIST]     Koepke, G., Young, W., Ladbury, J., and J. Coder,
              "Interference and Coexistence of Wireless Systems in
              Critical Infrastructure", National Institute of Standards
              and Technology report, DOI 10.6028/nist.tn.1885, July
              2015.

   [SEMINT]   Feeney, L., "Exploring semantic interference in
              heterogeneous sensor networks", Proceeding of the 1st ACM
              international workshop on Heterogeneous sensor and actor
              networks - HeterSanet '08 , DOI 10.1145/1374699.1374708,
              2008.

   [SURVEY]   Han, Y., Ekici, E., Kremo, H., and O. Altintas, "Spectrum
              sharing methods for the coexistence of multiple RF
              systems: A survey", Ad Hoc Networks Vol. 53, pp. 53-78,
              DOI 10.1016/j.adhoc.2016.09.009, December 2016.

Acknowledgements

   The authors would like to thank Michael Frey, Charalampos Orfanidis,
   Martin Jacobsson, and Per Gunningberg for their valuable
   collaboration in simulation and measurement studies of inter-network
   interference.  We would also like to thank Carsten Bormann for his
   support and encouragement in preparing this document, particularly
   the discussion of security considerations.

Authors' Addresses

   Laura Marie Feeney
   Uppsala University
   Box 337
   Uppsala  SE-751 05
   Sweden

   Email: lmfeeney@it.uu.se

Feeney & Fodor           Expires January 4, 2018               [Page 11]
Internet-Draft          inter-network coexistence              July 2017

   Viktoria Fodor
   KTH
   Osquldas vaeg 10
   Stockholm  SE-100 44
   Sweden

   Email: vjfodor@kth.se

Feeney & Fodor           Expires January 4, 2018               [Page 12]