RFC 2501






Network Working Group                                          S. Corson
Request for Comments: 2501                        University of Maryland
Category: Informational                                        J. Macker
                                               Naval Research Laboratory
                                                            January 1999


                   Mobile Ad hoc Networking (MANET):
   Routing Protocol Performance Issues and Evaluation Considerations

Status of this Memo



   This memo provides information for the Internet community.  It does
   not specify an Internet standard of any kind.  Distribution of this
   memo is unlimited.

Copyright Notice



   Copyright (C) The Internet Society (1999).  All Rights Reserved.

Abstract



   This memo first describes the characteristics of Mobile Ad hoc
   Networks (MANETs), and their idiosyncrasies with respect to
   traditional, hardwired packet networks.  It then discusses the effect
   these differences have on the design and evaluation of network
   control protocols with an emphasis on routing performance evaluation
   considerations.

1. Introduction



   With recent performance advancements in computer and wireless
   communications technologies, advanced mobile wireless computing is
   expected to see increasingly widespread use and application, much of
   which will involve the use of the Internet Protocol (IP) suite. The
   vision of mobile ad hoc networking is to support robust and efficient
   operation in mobile wireless networks by incorporating routing
   functionality into mobile nodes.  Such networks are envisioned to
   have dynamic, sometimes rapidly-changing, random, multihop topologies
   which are likely composed of relatively bandwidth-constrained
   wireless links.

   Within the Internet community, routing support for mobile hosts is
   presently being formulated as "mobile IP" technology.  This is a
   technology to support nomadic host "roaming", where a roaming host
   may be connected through various means to the Internet other than its
   well known fixed-address domain space. The host may be directly
   physically connected to the fixed network on a foreign subnet, or be



Corson & Macker              Informational                      [Page 1]

RFC 2501                MANET Performance Issues            January 1999


   connected via a wireless link, dial-up line, etc.  Supporting this
   form of host mobility (or nomadicity) requires address management,
   protocol interoperability enhancements and the like, but core network
   functions such as hop-by-hop routing still presently rely upon pre-
   existing routing protocols operating within the fixed network. In
   contrast, the goal of mobile ad hoc networking is to extend mobility
   into the realm of autonomous, mobile, wireless domains, where a set
   of nodes--which may be combined routers and hosts--themselves form
   the network routing infrastructure in an ad hoc fashion.

2. Applications



   The technology of Mobile Ad hoc Networking is somewhat synonymous
   with Mobile Packet Radio Networking (a term coined via during early
   military research in the 70's and 80's), Mobile Mesh Networking (a
   term that appeared in an article in The Economist regarding the
   structure of future military networks) and Mobile, Multihop, Wireless
   Networking (perhaps the most accurate term, although a bit
   cumbersome).

   There is current and future need for dynamic ad hoc networking
   technology.  The emerging field of mobile and nomadic computing, with
   its current emphasis on mobile IP operation, should gradually broaden
   and require highly-adaptive mobile networking technology to
   effectively manage multihop, ad hoc network clusters which can
   operate autonomously or, more than likely, be attached at some
   point(s) to the fixed Internet.

   Some applications of MANET technology could include industrial and
   commercial applications involving cooperative mobile data exchange.
   In addition,  mesh-based mobile networks can be operated as robust,
   inexpensive alternatives or enhancements to cell-based mobile network
   infrastructures. There are also existing and future military
   networking requirements for robust, IP-compliant data services within
   mobile wireless communication networks [1]--many of these networks
   consist of highly-dynamic autonomous topology segments. Also, the
   developing technologies of "wearable" computing and communications
   may provide applications for MANET technology. When properly combined
   with satellite-based information delivery, MANET technology can
   provide an extremely flexible method for establishing communications
   for fire/safety/rescue operations or other scenarios requiring
   rapidly-deployable communications with survivable, efficient dynamic
   networking. There are likely other applications for MANET technology
   which are not presently realized or envisioned by the authors.  It
   is, simply put, improved IP-based networking technology for dynamic,
   autonomous wireless networks.





Corson & Macker              Informational                      [Page 2]

RFC 2501                MANET Performance Issues            January 1999


3. Characteristics of MANETs



   A MANET consists of mobile platforms (e.g., a router with multiple
   hosts and wireless communications devices)--herein simply referred to
   as "nodes"--which are free to move about arbitrarily. The nodes may
   be located in or on airplanes, ships, trucks, cars, perhaps even on
   people or very small devices, and there may be multiple hosts per
   router. A MANET is an autonomous system of mobile nodes.  The system
   may operate in isolation, or may have gateways to and interface with
   a fixed network. In the latter operational mode, it is typically
   envisioned to operate as a "stub" network connecting to a fixed
   internetwork.  Stub networks carry traffic originating at and/or
   destined for internal nodes, but do not permit exogenous traffic to
   "transit" through the stub network.

   MANET nodes are equipped with wireless transmitters and receivers
   using antennas which may be omnidirectional (broadcast), highly-
   directional (point-to-point), possibly steerable, or some combination
   thereof. At a given point in time, depending on the nodes' positions
   and their transmitter and receiver coverage patterns, transmission
   power levels and co-channel interference levels, a wireless
   connectivity in the form of a random, multihop graph or "ad hoc"
   network exists between the nodes.  This ad hoc topology may change
   with time as the nodes move or adjust their transmission and
   reception parameters.

   MANETs have several salient characteristics:

      1) Dynamic topologies: Nodes are free to move arbitrarily; thus,
      the network topology--which is typically multihop--may change
      randomly and rapidly at unpredictable times, and may consist of
      both bidirectional and unidirectional links.

      2) Bandwidth-constrained, variable capacity links: Wireless links
      will continue to have significantly lower capacity than their
      hardwired counterparts. In addition, the realized throughput of
      wireless communications--after accounting for the effects of
      multiple access, fading, noise, and interference conditions,
      etc.--is often much less than a radio's maximum transmission rate.

      One effect of the relatively low to moderate link capacities is
      that congestion is typically the norm rather than the exception,
      i.e.  aggregate application demand will likely approach or exceed
      network capacity frequently. As the mobile network is often simply
      an extension of the fixed network infrastructure, mobile ad hoc
      users will demand similar services. These demands will continue to
      increase as multimedia computing and collaborative networking
      applications rise.



Corson & Macker              Informational                      [Page 3]

RFC 2501                MANET Performance Issues            January 1999


      3) Energy-constrained operation: Some or all of the nodes in a
      MANET may rely on batteries or other exhaustible means for their
      energy. For these nodes, the most important system design criteria
      for optimization may be energy conservation.

      4) Limited physical security: Mobile wireless networks are
      generally more prone to physical security threats than are fixed-
      cable nets.  The increased possibility of eavesdropping, spoofing,
      and denial-of-service attacks should be carefully considered.
      Existing link security techniques are often applied within
      wireless networks to reduce security threats. As a benefit, the
      decentralized nature of network control in MANETs provides
      additional robustness against the single points of failure of more
      centralized approaches.

   In addition, some envisioned networks (e.g. mobile military networks
   or highway networks) may be relatively large (e.g. tens or hundreds
   of nodes per routing area).  The need for scalability is not unique
   to MANETS. However, in light of the preceding characteristics, the
   mechanisms required to achieve scalability likely are.

   These characteristics create a set of underlying assumptions and
   performance concerns for protocol design which extend beyond those
   guiding the design of routing within the higher-speed, semi-static
   topology of the fixed Internet.

4. Goals of IETF Mobile Ad Hoc Network (manet) Working Group



   The intent of the newly formed IETF manet working group is to develop
   a peer-to-peer mobile routing capability in a purely mobile, wireless
   domain.  This capability will exist beyond the fixed network (as
   supported by traditional IP networking) and beyond the one-hop fringe
   of the fixed network.

   The near-term goal of the manet working group is to standardize one
   (or more) intra-domain unicast routing protocol(s), and related
   network-layer support technology which:

      * provides for effective operation over a wide range of mobile
      networking "contexts" (a context is a set of characteristics
      describing a mobile network and its environment);

      * supports traditional, connectionless IP service;

      * reacts efficiently to topological changes and traffic demands
      while maintaining effective routing in a mobile networking
      context.




Corson & Macker              Informational                      [Page 4]

RFC 2501                MANET Performance Issues            January 1999


   The working group will also consider issues pertaining to addressing,
   security, and interaction/interfacing with lower and upper layer
   protocols. In the longer term, the group may look at the issues of
   layering more advanced mobility services on top of the initial
   unicast routing developed.  These longer term issues will likely
   include investigating multicast and QoS extensions for a dynamic,
   mobile area.

5. IP-Layer Mobile Routing



   An improved mobile routing capability at the IP layer can provide a
   benefit similar to the intention of the original Internet, viz. "an
   interoperable internetworking capability over a heterogeneous
   networking infrastructure". In this case, the infrastructure is
   wireless, rather than hardwired, consisting of multiple wireless
   technologies, channel access protocols, etc.  Improved IP routing and
   related networking services provide the glue to preserve the
   integrity of the mobile internetwork segment in this more dynamic
   environment.

   In other words, a real benefit to using IP-level routing in a MANET
   is to provide network-level consistency for multihop networks
   composed of nodes using a *mixture* of physical-layer media; i.e. a
   mixture of what are commonly thought of as subnet technologies.  A
   MANET node principally consists of a router, which may be physically
   attached to multiple IP hosts (or IP-addressable devices), which has
   potentially *multiple* wireless interfaces--each interface using a
   *different* wireless technology.  Thus, a MANET node with interfaces
   using technologies A and B can communicate with any other MANET node
   possessing an interface with technology A or B.  The multihop
   connectivity of technology A forms a physical-layer multihop
   topology, the multihop connectivity of technology B forms *another*
   physical-layer topology (which may differ from that of A's topology),
   and the *union* of these topologies forms another topology (in graph
   theoretic terms--a multigraph), termed the "IP routing fabric", of
   the MANET.  MANET nodes making routing decisions using the IP fabric
   can intercommunicate using either or both physical-layer topologies
   simultaneously.  As new physical-layer technologies are developed,
   new device drivers can be written and another physical-layer multihop
   topology can be seamlessly added to the IP fabric.  Likewise, older
   technologies can easily be dropped.  Such is the functionality and
   architectural flexibility that IP-layer routing can support, which
   brings with it hardware economies of scale.

   The concept of a "node identifier" (separate and apart from the
   concept of an "interface identifier") is crucial to supporting the
   multigraph topology of the routing fabric. It is what *unifies* a set
   of wireless interfaces and identifies them as belonging to the same



Corson & Macker              Informational                      [Page 5]

RFC 2501                MANET Performance Issues            January 1999


   mobile platform.  This approach permits maximum flexibility in
   address assignment.  Node identifiers are used at the IP layer for
   routing computations.

5.1. Interaction with Standard IP Routing



   In the near term, it is currently envisioned that MANETs will
   function as *stub* networks, meaning that all traffic carried by
   MANET nodes will either be sourced or sinked within the MANET.
   Because of bandwidth and possibly power constraints, MANETs are not
   presently envisioned to function as *transit* networks carrying
   traffic which enters and then leaves the MANET (although this
   restriction may be removed by subsequent technology advances).  This
   substantially reduces the amount of route advertisement required for
   interoperation with the existing fixed Internet. For stub operation,
   routing interoperability in the near term may be achieved using some
   combination of mechanisms such as MANET-based anycast and mobile IP.
   Future interoperability may be achieved using mechanisms other than
   mobile IP.

   Interaction with Standard IP Routing will be greatly facilitated by
   usage of a common MANET addressing approach by all MANET routing
   protocols. Development of such an approach is underway which permits
   routing through a multi-technology fabric, permits multiple hosts per
   router and ensures long-term interoperability through adherence to
   the IP addressing architecture.  Supporting these features appears
   only to require identifying host and router interfaces with IP
   addresses, identifying a router with a separate Router ID, and
   permitting routers to have multiple wired and wireless interfaces.

6. MANET Routing Protocol Performance Issues



   To judge the merit of a routing protocol, one needs metrics--both
   qualitative and quantitative--with which to measure its suitability
   and performance.  These metrics should be *independent* of any given
   routing protocol.

   The following is a list of desirable qualitative properties of MANET
   routing protocols:

      1) Distributed operation:  This is an essential property, but it
      should be stated nonetheless.

      2) Loop-freedom:  Not required per se in light of certain
      quantitative measures (i.e. performance criteria), but generally
      desirable to avoid problems such as worst-case phenomena, e.g. a
      small fraction of packets spinning around in the network for
      arbitrary time periods.  Ad hoc solutions such as TTL values can



Corson & Macker              Informational                      [Page 6]

RFC 2501                MANET Performance Issues            January 1999


      bound the problem, but a more structured and well-formed approach
      is generally desirable as it usually leads to better overall
      performance.

      3) Demand-based operation:  Instead of assuming an uniform traffic
      distribution within the network (and maintaining routing between
      all nodes at all times), let the routing algorithm adapt to the
      traffic pattern on a demand or need basis.  If this is done
      intelligently, it can utilize network energy and bandwidth
      resources more efficiently, at the cost of increased route
      discovery delay.

      4) Proactive operation:  The flip-side of demand-based operation.
      In certain contexts, the additional latency demand-based operation
      incurs may be unacceptable.  If bandwidth and energy resources
      permit, proactive operation is desirable in these contexts.

      5) Security: Without some form of network-level or link-layer
      security, a MANET routing protocol is vulnerable to many forms of
      attack.  It may be relatively simple to snoop network traffic,
      replay transmissions, manipulate packet headers, and redirect
      routing messages, within a wireless network without appropriate
      security provisions. While these concerns exist within wired
      infrastructures and routing protocols as well, maintaining the
      "physical" security of of the transmission media is harder in
      practice with MANETs. Sufficient security protection to prohibit
      disruption of modification of protocol operation is desired. This
      may be somewhat orthogonal to any particular routing protocol
      approach, e.g. through the application of IP Security techniques.

      6) "Sleep" period operation:  As a result of energy conservation,
      or some other need to be inactive, nodes of a MANET may stop
      transmitting and/or receiving (even receiving requires power) for
      arbitrary time periods.  A routing protocol should be able to
      accommodate such sleep periods without overly adverse
      consequences. This property may require close coupling with the
      link-layer protocol through a standardized interface.

      7) Unidirectional link support:  Bidirectional links are typically
      assumed in the design of routing algorithms, and many algorithms
      are incapable of functioning properly over unidirectional links.
      Nevertheless, unidirectional links can and do occur in wireless
      networks. Oftentimes, a sufficient number of duplex links exist so
      that usage of unidirectional links is of limited added value.
      However, in situations where a pair of unidirectional links (in
      opposite directions) form the only bidirectional connection
      between two ad hoc regions, the ability to make use of them is
      valuable.



Corson & Macker              Informational                      [Page 7]

RFC 2501                MANET Performance Issues            January 1999


   The following is a list of quantitative metrics that can be used to
   assess the performance of any routing protocol.

      1) End-to-end data throughput and delay: Statistical measures of
      data routing performance (e.g., means, variances, distributions)
      are important. These are the measures of a routing policy's
      effectiveness--how well it does its job--as measured from the
      *external* perspective of other policies that make use of routing.

      2) Route Acquisition Time: A particular form of *external* end-
      to-end delay measurement--of particular concern with "on demand"
      routing algorithms--is the time required to establish route(s)
      when requested.

      3) Percentage Out-of-Order Delivery: An external measure of
      connectionless routing performance of particular interest to
      transport layer protocols such as TCP which prefer in-order
      delivery.

      4) Efficiency:  If data routing effectiveness is the external
      measure of a policy's performance, efficiency is the *internal*
      measure of its effectiveness.  To achieve a given level of data
      routing performance, two different policies can expend differing
      amounts of overhead, depending on their internal efficiency.
      Protocol efficiency may or may not directly affect data routing
      performance.  If control and data traffic must share the same
      channel, and the channel's capacity is limited, then excessive
      control traffic often impacts data routing performance.

      It is useful to track several ratios that illuminate the
      *internal* efficiency of a protocol in doing its job (there may be
      others that the authors have not considered):

         * Average number of data bits transmitted/data bit delivered--
         this can be thought of as a measure of the bit efficiency of
         delivering data within the network.  Indirectly, it also gives
         the average hop count taken by data packets.

         * Average number of control bits transmitted/data bit
         delivered--this measures the bit efficiency of the protocol in
         expending control overhead to delivery data.  Note that this
         should include not only the bits in the routing control
         packets, but also the bits in the header of the data packets.
         In other words, anything that is not data is control overhead,
         and should be counted in the control portion of the algorithm.






Corson & Macker              Informational                      [Page 8]

RFC 2501                MANET Performance Issues            January 1999


         * Average number of control and data packets transmitted/data
         packet delivered--rather than measuring pure algorithmic
         efficiency in terms of bit count, this measure tries to capture
         a protocol's channel access efficiency, as the cost of channel
         access is high in contention-based link layers.

   Also, we must consider the networking *context* in which a protocol's
   performance is measured.  Essential parameters that should be varied
   include:

      1) Network size--measured in the number of nodes

      2) Network connectivity--the average degree of a node (i.e. the
      average number of neighbors of a node)

      3) Topological rate of change--the speed with which a network's
      topology is changing

      4) Link capacity--effective link speed measured in bits/second,
      after accounting for losses due to multiple access, coding,
      framing, etc.

      5) Fraction of unidirectional links--how effectively does a
      protocol perform as a function of the presence of unidirectional
      links?

      6) Traffic patterns--how effective is a protocol in adapting to
      non-uniform or bursty traffic patterns?

      7) Mobility--when, and under what circumstances, is temporal and
      spatial topological correlation relevant to the performance of a
      routing protocol?  In these cases, what is the most appropriate
      model for simulating node mobility in a MANET?

      8) Fraction and frequency of sleeping nodes--how does a protocol
      perform in the presence of sleeping and awakening nodes?

   A MANET protocol should function effectively over a wide range of
   networking contexts--from small, collaborative, ad hoc groups to
   larger mobile, multihop networks.  The preceding discussion of
   characteristics and evaluation metrics somewhat differentiate MANETs
   from traditional, hardwired, multihop networks.  The wireless
   networking environment is one of scarcity rather than abundance,
   wherein bandwidth is relatively limited, and energy may be as well.

   In summary, the networking opportunities for MANETs are intriguing
   and the engineering tradeoffs are many and challenging.  A diverse
   set of performance issues requires new protocols for network control.



Corson & Macker              Informational                      [Page 9]

RFC 2501                MANET Performance Issues            January 1999


   A question which arises is "how should the *goodness* of a policy be
   measured?". To help answer that, we proposed here an outline of
   protocol evaluation issues that highlight performance metrics that
   can help promote meaningful comparisons and assessments of protocol
   performance.  It should be recognized that a routing protocol tends
   to be well-suited for particular network contexts, and less well-
   suited for others. In putting forth a description of a protocol, both
   its *advantages* and *limitations* should be mentioned so that the
   appropriate networking context(s) for its usage can be identified.
   These attributes of a protocol can typically be expressed
   *qualitatively*, e.g., whether the protocol can or cannot support
   shortest-path routing.  Qualitative descriptions of this nature
   permit broad classification of protocols, and form a basis for more
   detailed *quantitative* assessments of protocol performance. In
   future documents, the group may put forth candidate recommendations
   regarding protocol design for MANETs. The metrics and the philosophy
   presented within this document are expected to continue to evolve as
   MANET technology and related efforts mature.

7. Security Considerations



   Mobile wireless networks are generally more prone to physical
   security threats than are fixed, hardwired networks. Existing link-
   level security techniques (e.g. encryption) are often applied within
   wireless networks to reduce these threats.  Absent link-level
   encryption, at the network layer, the most pressing issue is one of
   inter-router authentication prior to the exchange of network control
   information.  Several levels of authentication ranging from no
   security (always an option) and simple shared-key approaches, to full
   public key infrastructure-based authentication mechanisms will be
   explored by the group.  As an adjunct to the working groups efforts,
   several optional authentication modes may be standardized for use in
   MANETs.

8. References



   [1] Adamson, B., "Tactical Radio Frequency Communication Requirements
       for IPng", RFC 1677, August 1994.













Corson & Macker              Informational                     [Page 10]

RFC 2501                MANET Performance Issues            January 1999


Authors' Addresses



   M. Scott Corson
   Institute for Systems Research
   University of Maryland
   College Park, MD 20742



   Phone: (301) 405-6630
   EMail: corson@isr.umd.edu


   Joseph Macker
   Information Technology Division
   Naval Research Laboratory
   Washington, DC 20375

   Phone: (202) 767-2001
   EMail: macker@itd.nrl.navy.mil

































Corson & Macker              Informational                     [Page 11]

RFC 2501                MANET Performance Issues            January 1999


Full Copyright Statement



   Copyright (C) The Internet Society (1999).  All Rights Reserved.

   This document and translations of it may be copied and furnished to
   others, and derivative works that comment on or otherwise explain it
   or assist in its implementation may be prepared, copied, published
   and distributed, in whole or in part, without restriction of any
   kind, provided that the above copyright notice and this paragraph are
   included on all such copies and derivative works.  However, this
   document itself may not be modified in any way, such as by removing
   the copyright notice or references to the Internet Society or other
   Internet organizations, except as needed for the purpose of
   developing Internet standards in which case the procedures for
   copyrights defined in the Internet Standards process must be
   followed, or as required to translate it into languages other than
   English.

   The limited permissions granted above are perpetual and will not be
   revoked by the Internet Society or its successors or assigns.

   This document and the information contained herein is provided on an
   "AS IS" basis and THE INTERNET SOCIETY AND THE INTERNET ENGINEERING
   TASK FORCE DISCLAIMS ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING
   BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION
   HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF
   MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE.
























Corson & Macker              Informational                     [Page 12]