Network Working Group F. Teraoka

Network Working Group F. Teraoka
Request for Comments: 5184 K. Gogo
Category: Experimental K. Mitsuya
R. Shibui
K. Mitani
KEIO University
May 2008

Unified Layer 2 (L2) Abstractions
for Layer 3 (L3)-Driven Fast Handover

Status of This Memo

This memo defines an Experimental Protocol for the Internet
community. It does not specify an Internet standard of any kind.
Discussion and suggestions for improvement are requested.
Distribution of this memo is unlimited.

IESG Note

This document is not an IETF Internet Standard. It represents the
consensus of the MOBOPTS Research Group of the Internet Research Task
Force. It may be considered for standardization by the IETF in the
future.

Abstract

This document proposes unified Layer 2 (L2) abstractions for Layer 3
(L3)-driven fast handovers. For efficient network communication, it
is vital for a protocol layer to know or utilize other layers'
information, such as the form of L2 triggers. However, each protocol
layer is basically designed independently. Since each protocol layer
is also implemented independently in current operating systems, it is
very hard to exchange control information between protocol layers.
This document defines nine kinds of L2 abstractions in the form of
"primitives" to achieve fast handovers in the network layer as a
means of solving the problem. This mechanism is called "L3-driven
fast handovers" because the network layer initiates L2 and L3
handovers by using the primitives. This document is a product of the
IP Mobility Optimizations (MobOpts) Research Group.

Teraoka, et al. Experimental [Page 1]

RFC 5184 L2 Abstractions for L3-Driven Fast Handover May 2008

Table of Contents

1. Introduction ....................................................3
2. Terminology .....................................................3
3. Primitives for L2 Abstractions ..................................4
4. Definitions of Primitives .......................................6
4.1. L2-LinkStatus (Type 1) .....................................6
4.2. L2-PoAList (Type 1) ........................................6
4.3. L2-PoAFound (Type 2) .......................................6
4.4. L2-PoALost (Type 2) ........................................6
4.5. L2-LinkUp (Type 2) .........................................7
4.6. L2-LinkDown (Type 2) .......................................7
4.7. L2-LinkStatusChanged (Type 2) ..............................7
4.8. L2-LinkConnect (Type 3) ....................................7
4.9. L2-LinkDisconnect (Type 3) .................................8
5. Definitions of Static Parameters ................................8
5.1. Network Interface ID .......................................8
6. Definitions of Dynamic Parameters ...............................8
6.1. PoA (Point of Attachment) ..................................8
6.2. Condition ..................................................8
6.3. PoA List ...................................................9
6.4. Enable/Disable .............................................9
6.5. Ack/Error ..................................................9
7. Architectural Considerations ....................................9
8. Security Considerations ........................................13
9. Acknowledgements ...............................................14
10. References ....................................................14
10.1. Normative References .....................................14
10.2. Informative References ...................................14
Appendix A. Example Scenario ....................................16
Appendix B. Example Operation for FMIPv6 ........................17
B.1. Example Operation-1 for FMIPv6 ............................18
B.2. Example Operation-2 for FMIPv6 ............................20
B.3. Experiment ................................................21
Appendix C. Example Mapping between L2 Primitives and
Primitives in IEEE 802.11 and IEEE 802.16e ..........22
Appendix D. Example Mapping of Primitives and IEEE 802.11 .......24
D.1. L2-LinkStatus ............................................24
D.2. L2-PoAList ................................................24
D.3. L2-PoAFound ..............................................24
D.4. L2-PoALost ................................................25
D.5. L2-LinkUp ................................................25
D.6. L2-LinkDown ..............................................25
D.7. L2-LinkStatusChanged ......................................25
D.8. L2-LinkConnect ............................................26
D.9. L2-LinkDisconnect ........................................26
Appendix E. Implementation and Evaluation of the Proposed
Model ................................................26

Teraoka, et al. Experimental [Page 2]

RFC 5184 L2 Abstractions for L3-Driven Fast Handover May 2008

1. Introduction

Recent years have witnessed the rapid proliferation of wireless
networks as well as mobile devices accessing them. Unlike wired
network environments, wireless networks are characterized by
dynamically changing radio conditions, connectivity, and available
bandwidth. For efficient network communication, it is vital for a
protocol layer to know or utilize other layers' control information.
Mobile IPv4 [2] and Mobile IPv6 [3] have been standardized to support
communication with mobile nodes. There are several proposals for
seamless handovers in IPv6 networks, such as Fast Handovers for
Mobile IPv6 (FMIPv6) [4] and Hierarchical Mobile IPv6 (HMIPv6) [5].
In FMIPv6, the network layer must know in advance the indication of a
handover from the link layer to achieve seamless handovers. However,
control information exchange between protocol layers is typically not
available because each protocol layer is designed independently.

To solve the problem, this document defines nine kinds of L2
abstractions in the form of "primitives" to achieve fast handovers in
the network layer. This mechanism is called "L3-driven fast
handovers" because the network layer initiates L2 and L3 handovers by
using the primitives.

IEEE 802.21 [6] also defines several services that make use of L2
information. For the sake of ease of implementation and deployment,
the primitives defined in this document make use of only the
information available in the mobile node, while IEEE 802.21 [6]
introduces the information server in the network to provide the
mobile node with network-related information, such as a global
network map.

This document represents the consensus of the MobOpts Research Group.
It has been reviewed by Research Group members active in the specific
area of work.

2. Terminology

This document uses the following terms:

L3-Driven Fast Handover

The handover mechanism that is initiated by the network layer on a
mobile node. Since this mechanism allows handover preparation in
L3 before the start of an L2 handover on the mobile node, it can
reduce packet loss during a handover. The L3-driven fast handover
mechanism requires L2 information as a trigger for a handover
procedure.

Teraoka, et al. Experimental [Page 3]

RFC 5184 L2 Abstractions for L3-Driven Fast Handover May 2008

PoA

The point of attachment of a mobile node (e.g., an access point in
IEEE 802.11 networks [7]).

Primitive

A unit of information that is sent from one layer to another.
There are four classes of primitives: Request, Confirm,
Indication, and Response. One or more classes of a primitive are
exchanged, depending on the type of primitive.

The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
"SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
document are to be interpreted as described in RFC 2119 [1].

3. Primitives for L2 Abstractions

Each layer offers its services in the form of primitives. Four
classes of primitives are defined, as shown in Figure 1. Request is
issued by the layer that wants to get the services or information
from another layer, and Confirm is the acknowledgment of the request.
Indication is the notification of the information to the layer that
requested the service, and Response is the acknowledgment of the
indication. In this architecture, communication between layers is
symmetrical.

------------------------- -----------------------------
Request Response
|| /\ /\ ||
Layer N || || || ||
------------||------||--- -------||------||------------
|| || || ||
\/ || || \/
Layer N-m Confirm Indication
------------------------- -----------------------------

Figure 1: Interaction Model between Layers

The primitive consists of five fields: protocol layer ID, protocol
ID, primitive class (Request, Response, Indication, or Confirm),
primitive name, and parameters. The protocol layer ID specifies to
which layer this primitive should be sent, e.g., Layer 2 or Layer 3.
The protocol ID specifies to which protocol entity this primitive
should be sent, e.g., IEEE 802.11 [7] or IEEE 802.3 [8].

Teraoka, et al. Experimental [Page 4]

RFC 5184 L2 Abstractions for L3-Driven Fast Handover May 2008

For unified L2 abstractions for L3-driven fast handovers, three
different usages of primitives are defined, as described below:

Type 1. To provide L2 information to upper layers immediately

This type of primitive is used to provide the L2 information to
upper layers immediately. The Request and Confirm classes of
primitives MUST be exchanged for the interaction. The Request
primitive is for an acquisition request for the L2 information.
The Confirm primitive is for the answer.

Type 2. To notify upper layers of L2 events asynchronously

This type of primitive is used to notify upper layers of L2 events
asynchronously. The Request, Confirm, and Indication classes of
primitive MUST be exchanged, and the Response class MAY be
exchanged for the interaction. The Request and Confirm primitives
are used just for registration. When an event occurs, the
Indication primitive is asynchronously delivered to the upper
layer.

Type 3. To control L2 actions from upper layers

This type of primitive is used to control L2 actions from upper
layers. The Request and Confirm classes of primitives MUST be
exchanged for the interaction. The Request primitive is a request
for operation. Ack or Nack returns immediately as the Confirm
primitive.

A protocol entity can register primitives anytime by exchanging the
Request and Confirm messages that include the fields defined above.
When the registered event occurs, the Indication and Response
messages are exchanged as well.

The way to exchange a message between protocol entities is beyond the
scope of this document. Any information-exchange method between
layers, such as the work in [10], can be used.

The timing for sending an Indication primitive is also beyond the
scope of this document. For example, a layer 2 event is generated
when layer 2 status has been changed, and this depends upon how
scanning algorithms, for example, are implemented.

Teraoka, et al. Experimental [Page 5]

RFC 5184 L2 Abstractions for L3-Driven Fast Handover May 2008

4. Definitions of Primitives

To obtain and exchange L2 information, the following primitives are
defined. Appendix C shows example mapping between the L2 primitives
and the primitives in IEEE 802.11 [7] and IEEE 802.16e [9].

4.1. L2-LinkStatus (Type 1)

The L2-LinkStatus.request primitive is sent to the link layer when an
upper layer requires the current information of a link. The
L2-LinkStatus.request primitive contains the "Network Interface ID"
parameter (see Section 5.1). In response, the L2-LinkStatus.confirm
primitive returns. The L2-LinkStatus.confirm primitive contains
three parameters: "Network Interface ID", "PoA", and "Condition".
"PoA" and "Condition" indicate the current status of the link between
the mobile node and a PoA.

4.2. L2-PoAList (Type 1)

The L2-PoAList.request primitive is sent to the link layer when an
upper layer requires a list of the candidate PoAs. The
L2-PoAList.request primitive contains the "Network Interface ID"
parameter. In response, the L2-PoAList.confirm primitive returns the
"Network Interface ID" parameter and the "PoA List" parameter. The
"PoA List" parameter is a list of the candidate PoAs.

4.3. L2-PoAFound (Type 2)

The L2-PoAFound.indication primitive is asynchronously provided to an
upper layer when new PoAs are detected. This primitive carries the
"Network Interface ID" parameter and the "PoA List" parameter. The
"PoA List" parameter contains information on new PoAs detected by the
mobile node. In order to use this notification, the registration
process using the L2-PoAFound.request primitive and the
L2-PoAFound.confirm primitive is needed in advance. The
L2-PoAFound.request primitive has two parameters: "Network Interface
ID" and "Enable/Disable". The "Enable/Disable" parameter shows
whether this notification function is turned on. When this
registration succeeds, the L2-PoAFound.confirm primitive returns with
the "Network Interface ID" parameter and the "Ack" parameter in
response.

4.4. L2-PoALost (Type 2)

The L2-PoALost.indication primitive is asynchronously provided to an
upper layer when a PoA included in the list of candidate PoAs
disappears. This primitive carries the "Network Interface ID"
parameter and the "PoA List" parameter. The "PoA List" parameter

Teraoka, et al. Experimental [Page 6]

RFC 5184 L2 Abstractions for L3-Driven Fast Handover May 2008

contains information on the PoAs that disappeared from the list of
candidates. The registration process using the L2-PoALost.request
primitive and the L2-PoALost.confirm primitive is similar to the
L2-PoAFound primitive described above.

4.5. L2-LinkUp (Type 2)

The L2-LinkUp.indication primitive is asynchronously provided to an
upper layer when a new link is connected and IP packets can be
transmitted through the new link. As described in RFC 4957 [12],
what "link is connected" means depends on link types. For example,
in case of the infrastructure mode in IEEE 802.11 [7] (WiFi), this
primitive is provided when an association to an access point is
established. This primitive carries the "Network Interface ID"
parameter and the "PoA" parameter. The L2-LinkUp.request primitive
contains the "Network Interface ID" parameter and the
"Enable/Disable" parameter for registration. When the registration
succeeds, the L2-LinkUp.confirm primitive with the "Network Interface
ID" parameter and the "Ack" parameter returns.

4.6. L2-LinkDown (Type 2)

The L2-LinkDown.indication primitive is asynchronously provided to an
upper layer when an existing link is disconnected and IP packets
cannot be transmitted through the link. The registration processing
is the same as the L2-LinkUp primitive described above.

4.7. L2-LinkStatusChanged (Type 2)

The L2-LinkStatusChanged.indication primitive is asynchronously
provided to an upper layer when the status of a link has changed.
This notification contains three parameters: "Network Interface ID",
"PoA", and "Condition". The "PoA" parameter indicates the attachment
point at which the link quality changed. In the registration
processing, the L2-LinkStatusChanged.request primitive carries the
"Network Interface ID" parameter, the "Enable/Disable" parameter, and
the "Condition" parameter. "Condition" indicates the event type and
the threshold for the Indication.

4.8. L2-LinkConnect (Type 3)

The L2-LinkConnect.request primitive is sent to the link layer when
an upper layer has to establish a new link to the specific "PoA".
This primitive carries the "Network Interface ID" parameter and the
"PoA" parameter. This operation begins after the link layer returns
the L2-LinkConnect.confirm primitive with "Ack".

Teraoka, et al. Experimental [Page 7]

RFC 5184 L2 Abstractions for L3-Driven Fast Handover May 2008

4.9. L2-LinkDisconnect (Type 3)

The L2-LinkDisconnect.request primitive is sent to the link layer
when an upper layer has to tear down an existing link to the specific
"PoA". This primitive carries the "Network Interface ID" parameter
and the "PoA" parameter. This operation begins after the link layer
returns the L2-LinkDisconnect.confirm primitive with "Ack".

5. Definitions of Static Parameters

This section lists static parameters. Once the values of static
parameters are configured, they basically remain unchanged during
communication. The following parameters are transferred as a part of
primitives.

5.1. Network Interface ID

The "Network Interface ID" parameter uniquely identifies the network
interface in the node. The syntax of the identifier is
implementation-specific (e.g., name, index, or unique address
assigned to each interface). This parameter also contains the
network interface type that indicates the kind of technology of the
network interface (e.g., IEEE 802.11a/b/g [7], Third Generation
Partnership Project (3GPP), etc.). This parameter is required in all
primitives.

6. Definitions of Dynamic Parameters

This section lists dynamic parameters. The values of dynamic
parameters change frequently during communication. The following
parameters are transferred as a part of primitives.

6.1. PoA (Point of Attachment)

The "PoA" parameter uniquely identifies the PoA.

6.2. Condition

The "Condition" parameter consists of the following sub-parameters:
available bandwidth and link quality level. These sub-parameters are
the abstracted information that indicates the current quality of
service. The abstraction algorithms of sub-parameters depend on
hardware devices and software implementation. The useful range of
link quality is divided into five levels: EXCELLENT, GOOD, FAIR, BAD,
and NONE, in descending order. The quality levels of an L2 device

Teraoka, et al. Experimental [Page 8]

RFC 5184 L2 Abstractions for L3-Driven Fast Handover May 2008

are independent of those of other devices. However, making decisions
based on these metrics is error prone and not guaranteed to result in
an optimal choice of links. An example of the thresholds among the
five levels in IEEE 802.11 [7] is described in Appendix E.

6.3. PoA List

The "PoA List" parameter consists of arbitrary couples of two
sub-parameters: "PoA" and "Condition". This parameter shows a list
of PoAs and their conditions.

6.4. Enable/Disable

The "Enable/Disable" flag is used for turning event notification on/
off. When an upper layer needs notifications, the Request primitive
with "Enable" is sent to the link layer as registration. When an
upper layer needs no more notifications, the Request primitive with
"Disable" is sent.

6.5. Ack/Error

When an upper layer requests some notifications, the link layer
receives and confirms this Request. If the Request is valid, the
Confirm primitive with "Ack" is sent to the upper layer. If it is
invalid, the Confirm with "Error" is sent to the upper layer.

7. Architectural Considerations

RFC 4907 [11] discusses the role and the issues of link indications
within the Internet Architecture. This section discusses the
architectural considerations mentioned in Section 2 of RFC 4907.

1. Proposals should avoid use of simplified link models in
circumstances where they do not apply.

The information in each layer should be abstracted before it is
sent to another layer. For example, in IEEE 802.11 [7], the
Received Signal Strength Indicator (RSSI), the number of
retransmissions, and the existence of association between the
mobile node and the access point are used so that the link
layer indications can adjust themselves to various environments
or situations. The thresholds needed for some link indications
are defined from empirical study.

In the conventional protocol-layering model, the Protocol
Entity (PE) is defined as the entity that processes a specific
protocol. Our proposal introduced the Abstract Entity (AE) to

Teraoka, et al. Experimental [Page 9]

RFC 5184 L2 Abstractions for L3-Driven Fast Handover May 2008

achieve link independency of the link indications. An AE and a
PE make a pair. An AE abstracts the PE-dependent information
to the PE-independent information.

Figure 2 shows AEs and PEs using primitives.

2. Link indications should be clearly defined, so that it is
understood when they are generated on different link layers.

To make the link information/indications clear, our proposal
defines the 4 types of primitives: Request/Confirm and
Indication/Response, as described in Section 3. The Request is
used to obtain the information of another layer. The Confirm
is the reply to the request and it includes the requested
information. The Indication is generated when a particular
event occurs. The Response is the reply to the indication.

In our proposal on IEEE 802.11b [7], L2-LinkUp is defined as
the status in which an association to the Access Point (AP) is
established, and L2-LinkDown is defined as the status in which
an association to the AP is not established.
L2-LinkStatusChanged is generated when the link quality goes
below the predefined threshold. Since the Received Signal
Strength Indicator (RSSI) and the number of retransmissions are
used to abstract and evaluate the link quality, L2-
LinkStatusChanged represents the link quality in both
directions. It should use an average of the RSSI or the number
of retransmissions damped for one second or more to cope with
transient link conditions.

3. Proposals must demonstrate robustness against misleading
indications.

Since RSSI changes significantly even when the mobile node
stands still according to the measurements in our experiments,
our proposal uses the RSSI, the number of retransmissions, and
the existence of an association to calculate the link status,
as described above. In our experiments, there were some
"ping-pong" handovers between two APs. Such ping-pong
handovers could be reduced by detecting the most suitable AP by
L2-LinkStatus when L2-LinkStatusChanged is notified. The use
of L2 indications is related to parameter thresholds that
trigger handover. These thresholds vary based on the
deployment scenario and, if not configured properly, could lead
to misleading indications.

Teraoka, et al. Experimental [Page 10]

RFC 5184 L2 Abstractions for L3-Driven Fast Handover May 2008

4. Upper layers should utilize a timely recovery step so as to
limit the potential damage from link indications determined to
be invalid after they have been acted on.

The proposed L3-driven handover described in Appendix E uses
the L2-LinkStatusChanged indication as the trigger for starting
handover. L2-LinkStatusChanged is indicated when the link
quality goes below a specific threshold. This indication is
not canceled even if the link quality goes up soon. As
described above, L2-LinkStatus can be used to detect the most
suitable AP. The IP layer can cancel a handover if it finds
that the current AP is the most suitable one by using
L2-LinkStatus when L2-LinkStatusChanged is notified.

5. Proposals must demonstrate that effective congestion control is
maintained.

Since this mechanism is coupled to the IP layer, and not
directly to the transport layer, the proposed mechanism does
not directly affect congestion control.

6. Proposals must demonstrate the effectiveness of proposed
optimizations.

In IPv6 mobility, the L3-driven handover mechanism using link
indications can dramatically reduce gap time due to handover.
The L3-driven handover mechanism needs the L2-LinkStatusChanged
indication to predict disconnection. But L2-LinkStatusChanged
is not trusted sometimes because it is difficult to abstract
the link quality. Invalid L2-LinkStatusChanged may cause
redundant handover. A handover mechanism using only L2-LinkUp/
L2-LinkDown can also reduce gap time modestly. An example of
an implementation and evaluation of the L3-driven handover
mechanism is described in Appendix E.

7. Link indications should not be required by upper layers in
order to maintain link independence.

Our proposal does not require any modifications to the
transport and upper layers.

8. Proposals should avoid race conditions, which can occur where
link indications are utilized directly by multiple layers of
the stack.

Since our proposal defines the link indications only to the IP
layer, race conditions between multiple layers never occur.

Teraoka, et al. Experimental [Page 11]

RFC 5184 L2 Abstractions for L3-Driven Fast Handover May 2008

9. Proposals should avoid inconsistencies between link and routing
layer metrics.

Our proposal does not deal with routing metrics.

10. Overhead reduction schemes must avoid compromising
interoperability and introducing link-layer dependencies into
the Internet and transport layers.

As described above, the link indications in our proposal are
abstracted to the information independent of link types to
reduce the gap time due to a handover, and the ordinary host
can execute handover without using the link indications defined
in our proposal.

11. Proposals advocating the transport of link indications beyond
the local host need to carefully consider the layering,
security, and transport implications. In general, implicit
signals are preferred to explicit transport of link indications
since they add no new packets in times of network distress,
operate more reliably in the presence of middle boxes, such as
NA(P)Ts (Network Address (Port) Translations), are more likely
to be backward compatible, and are less likely to result in
security vulnerabilities.

Our proposal does not define the exchange of link indications
between nodes.

Teraoka, et al. Experimental [Page 12]

RFC 5184 L2 Abstractions for L3-Driven Fast Handover May 2008

---------------------------------------------------------
----------=========== ----------===========
| |[ ] | |[ ]
| PE |[ AE ] | PE |[ AE ]
| |[ ] | |[ ]
----------=========== ----------===========
Layer N || /\ || /\
------------||---||-------------------||---||------------
Request|| || Response|| ||
|| || || ||
|| || || ||
|| ||Confirm || ||Indication
------------||---||-------------------||---||------------
\/ || \/ ||
----------=========== ----------===========
| |[ ] | |[ ]
| PE |[ AE ] | PE |[ AE ]
| |[ ] | |[ ]
----------=========== ----------===========
Layer N-m
---------------------------------------------------------

Figure 2: AE and PE with Primitives

8. Security Considerations

RFC 4907 [11] discusses the role and issues of link indications
within the Internet Architecture. This section discusses the
security considerations mentioned in Section 4 of RFC 4907.

1. Spoofing

The proposed primitives suffer from spoofed link-layer control
frames. For example, if a malicious access point is set up and
spoofed beacon frames are transmitted, L2-PoAFound.indication
is generated in the mobile node. As a result, the mobile node
may establish an association with the malicious access point by
an L2-LinkConnect.request.

2. Indication validation

Transportation of the link indications between nodes is not
assumed; hence, this consideration is beyond the scope of our
proposal.

Teraoka, et al. Experimental [Page 13]

RFC 5184 L2 Abstractions for L3-Driven Fast Handover May 2008

3. Denial of service

Since this proposal does not change link-layer protocols, no
more insecurity is added to a particular link-layer protocol.
However, the proposed primitives suffer from denial-of-service
attacks by spoofed link-layer frames. For example, L2-
PoAFound.indication and L2-PoALost.indication may frequently be
generated alternately if a malicious access point frequently
transmits control frames that indicate strong RSSI and weak
RSSI alternately.

9. Acknowledgements

The authors gratefully acknowledge the contributions of Jukka Manner,
Christian Vogt, and John Levine for their review.

10. References

10.1. Normative References

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

10.2. Informative References

[2] Perkins, C., Ed., "IP Mobility Support for IPv4", RFC 3344,
August 2002.

[3] Johnson, D., Perkins, C., and J. Arkko, "Mobility Support in
IPv6", RFC 3775, June 2004.

[4] Koodli, R., Ed., "Fast Handovers for Mobile IPv6", RFC 4068,
July 2005.

[5] Soliman, H., Castelluccia, C., El Malki, K., and L. Bellier,
"Hierarchical Mobile IPv6 Mobility Management (HMIPv6)", RFC
4140, August 2005.

[6] "Draft IEEE Standard for Local and Metropolitan Area Networks:
Media Independent Handover Services", IEEE P802.21/D02.00,
September 2006.

[7] IEEE, "802.11-2007 IEEE Standard for LAN/MAN - Specific
requirements Part 11: Wireless LAN Medium Access Control (MAC)
and Physical Layer (PHY) Specifications", 2007.

Teraoka, et al. Experimental [Page 14]

RFC 5184 L2 Abstractions for L3-Driven Fast Handover May 2008

[8] IEEE, "802.3, 2000 EDITION ISO/IEC 8802-3:2000 (E) Information
Technology - LAN/MAN - Part 3: Carrier Sense Multiple Access
with Collision Detection (CSMA/CD) Access Method and Physical
Layer Specifications", 2000.

[9] IEEE, "802.16e-2005 & 802.16/COR1 Part 16: Amendment for
Physical & Medium Access Control Layers for Combined Fixed &
Mobile Operation", 2006.

[10] Gogo, K., Shibu, R., and F. Teraoka, "An L3-Driven Fast Handover
Mechanism in IPv6 Mobility", In Proc. of International Symposium
on Applications and the Internet (SAINT2006) Workshop in IPv6,
February 2006.

[11] Aboba, B., Ed., "Architectural Implications of Link
Indications", RFC 4907, June 2007.

[12] Krishnan, S., Ed., Montavont, N., Njedjou, E., Veerepalli, S.,
and A. Yegin, Ed., "Link-Layer Event Notifications for Detecting
Network Attachments", RFC 4957, August 2007.

[13] Ishiyama, M., Kunishi, M., Uehara, K., Esaki, H., and F.
Teraoka, "LINA: A New Approach to Mobility Support in Wide Area
Networks", IEICE Transactions on Communication vol. E84-B, no.
8, pp. 2076-2086, August 2001.

Teraoka, et al. Experimental [Page 15]

RFC 5184 L2 Abstractions for L3-Driven Fast Handover May 2008

Appendix A. Example Scenario

For example, the picture below shows L3-driven fast handover
mechanism using the L2 triggers on a mobile node (MN).

L2 L3
| |
|<----------LinkUP.req-----------|
|-----------LinkUP.cnf---------->|
|<-----LinkStatusChanged.req-----|
|------LinkStatusChanged.cnf---->|
= =
| |
Low |
Signal---LinkStatusChanged.ind---->|
| |
|<----------PoAList.req----------|
|-----------PoAList.cnf------>Handover
| Preparation
|<-------LinkConnect.req---------|
L2 Handover--LinkConnect.cnf-------->:
: :
: :
finish---------LinkUp.ind----->L3 Handover
| finish
| |

L2: Link Layer on MN
L3: Network Layer on MN
req: Request
cnf: Confirm
ind: Indication

Figure 3: L3-Driven Fast Handover Mechanism

First, L3 issues LinkUp.request to receive LinkUp.indication when the
link becomes available. L3 also issues LinkStatusChanged.request to
receive LinkStatusChanged.indication when the link quality goes below
the threshold.

In the beginning of the L3-driven handover procedure, L2 detects that
the radio signal strength is going down. Then, L2 sends
L2-LinkStatusChanged.indication to L3. L3 prepares for handover
(e.g., Care-of Address (CoA) generation, Duplicate Address Detection
(DAD), Neighbor Discovery (ND) cache creation, and routing table
setting) and sends L2-PoAList.request to L2 if the list of access
points is needed.

Teraoka, et al. Experimental [Page 16]

RFC 5184 L2 Abstractions for L3-Driven Fast Handover May 2008

If L3 decides to perform handover according to some rules, L3 sends
L2-LinkConnect.request with some parameters about candidate access
points to request L2 handover. L2 handover begins after L2 sends
L2-LinkConnect.confirm to L3. When the L2 handover finishes, L2
sends L2-LinkUp.indication to notify L3. Finally, L3 performs
handover (e.g., sending a Binding Update (BU)).

One of the important features of L3-driven fast handover using
primitives is that L3 handover preparation can be done during
communication. So, it can reduce disruption time during handover.

Appendix B. Example Operation for FMIPv6

There are two scenarios of L3-driven fast handover for FMIPv6.
Scenario 2 is different from scenario 1 for the timing of sending
some messages.

Teraoka, et al. Experimental [Page 17]

RFC 5184 L2 Abstractions for L3-Driven Fast Handover May 2008

B.1. Example Operation-1 for FMIPv6

Figure 4 shows the predictive mode of FMIPv6 operation with an
L3-driven link-switching mechanism.

MN-L2 MN-L3 PAR-L3
| | |
AP<----------PoAList.req----------| |
Scan----------PoAList.cnf--------->| |
| |---RtSolPr-->|
| |<--PrRtAdv---|
|----------PoAFound.ind--------->| |
| |---RtSolPr-->|
| |<--PrRtAdv---|
| | |
~ ~ ~
| | |
Low | |
Signal---LinkStatusChanged.ind---->| | NAR-L3
| |-----FBU---->| |
| | |----HI---->|
| | |<--HAck----|
| |<----FBack---| |
|<-------LinkConnect.req---L3 Handover | |
L2 Handover--LinkConnect.cnf-------->: |
: : |
: : |
finish---------LinkUp.ind---------->: |
| :-----------FNA---------->|
| finish<======packets=========|
| | |

MN-L2 : Link Layer on Mobile Node
MN-L3 : Network Layer on Mobile Node
PAR-L3 : Network Layer on Previous Access Router
NAR-L3 : Network Layer on New Access Router
req : Request
cnf : Confirm
ind : Indication
RtSolPr : Router Solicitation for Proxy
PrRtAdv : Proxy Router Advertisement
FBU : Fast Binding Update
FBack : Fast Binding Acknowledgment
FNA : Fast Neighbor Advertisement
HI : Handover Initiate
HAck : Handover Acknowledge

Figure 4: L3-Driven Fast Handover Mechanism with FMIPv6 Scenario 1

Teraoka, et al. Experimental [Page 18]

RFC 5184 L2 Abstractions for L3-Driven Fast Handover May 2008

When MN establishes link connectivity to PAR, it performs router
discovery. MN sends an RtSolPr message to PAR to resolve the access
point identifiers to the subnet router information. To send RtSolPr,
MN discovers one or more access points by sending L2-PoAList.request
to the link layer. As a response to L2-PoAList.request,
L2-PoAList.confirm returns with "PoA List" parameter that contains
identifiers and conditions of nearby access points. After initial AP
discovery, L2-PoAFound.indication with "PoA List" is also sent from
the link layer when one or more access points are discovered.

When the link layer of MN detects that radio signal strength is
dropping, it sends L2-LinkStatusChanged.indication to the network
layer. Then, MN sends the FBU message to PAR as the beginning of the
L3 handover procedure. The NCoA required for the FBU message is
determined according to the MN's policy database and the received
PrRtAdv message. As a response to the FBU message, MN receives the
FBack message and then immediately switches its link by
L2-LinkConnect.request with the specific "PoA" parameter. The
handover policy of the MN is outside the scope of this document.

After L2 handover, the link layer of the MN sends
L2-LinkUp.indication to the network layer. MN immediately sends the
FNA message to the New Access Router (NAR). The NAR will send
tunneled and buffered packets to MN.

Teraoka, et al. Experimental [Page 19]

RFC 5184 L2 Abstractions for L3-Driven Fast Handover May 2008

B.2. Example Operation-2 for FMIPv6

Figure 5 shows the predictive mode of FMIPv6 operation with an
L3-driven link-switching mechanism.

MN-L2 MN-L3 PAR-L3
| | |
AP<----------PoAList.req----------| |
Scan----------PoAList.cnf--------->| |
| |---RtSolPr-->|
| |<--PrRtAdv---|
|----------PoAFound.ind--------->| |
| |---RtSolPr-->|
| |<--PrRtAdv---|
| | |
~ ~ ~
| | |
Low | |
Signal---LinkStatusChanged.ind---->| | NAR-L3
| |-----FBU---->| |
|<-------LinkConnect.req---L3 Handover | |
L2 Handover--LinkConnect.cnf-------->: | |
| | |----HI---->|
| | |<--HAck----|
| | <-FBack-|---FBack-->|
| |<----FBack---------------|
: : |
finish---------LinkUp.ind---------->: |
| :-----------FNA---------->|
| finish<======packets=========|
| | |

MN-L2 : Link Layer on Mobile Node
MN-L3 : Network Layer on Mobile Node
PAR-L3 : Network Layer on Previous Access Router
NAR-L3 : Network Layer on New Access Router
req : Request
cnf : Confirm
ind : Indication
RtSolPr : Router Solicitation for Proxy
PrRtAdv : Proxy Router Advertisement
FBU : Fast Binding Update
FBack : Fast Binding Acknowledgment
FNA : Fast Neighbor Advertisement
HI : Handover Initiate
HAck : Handover Acknowledge

Figure 5: L3-Driven Fast Handover Mechanism with FMIPv6 Scenario 2

Teraoka, et al. Experimental [Page 20]

RFC 5184 L2 Abstractions for L3-Driven Fast Handover May 2008

Unlike scenario 1, MN in scenario 2 sends LinkConnect.req from the
network layer to the link layer after MN sends the FBU message. As
PAR sends the FBack messages not only to PAR's subnet but also to
NAR's subnet, MN can get the FBack message sent by PAR through NAR,
and then it moves to NAR.

B.3. Experiment

We implemented FMIPv6 and the proposed L2 primitives on FreeBSD-5.4.
Figure 6 shows our test network. MN is connected to access routers
by using IEEE802.11a wireless LAN. MN moves from PAR to NAR.

|
+--+---+
|Router|
+--+---+
| 100BaseTX
---+--------+---------+---------+---------+------------
| | | |
+--+--+ +--+--+ +--+--+ +--+--+
| PAR | | NAR | | HA | | CN |
+-----+ +-----+ +-----+ +-----+
| |
IEEE802.11a IEEE802.11a PAR, NAR: nexcom EBC
|Channel 7 |Channel7 MN: ThinkPad X31
OS: FreeBSD-5.4
| | KAME/SHISA/TARZAN
+-----+ +-----+
| MN | -------> | MN |
+-----+ +-----+

Figure 6: Test Network

Scenario 1 was used in this experiment since it was more stable than
scenario 2. Upon receiving L2-LinkStatusChanged.indication, L3 of MN
sent the FBU message and then received the FBack message. It took
20ms from the transmission of the FBU message to the reception of the
FBack message. After receiving the FBack message, L3 of MN issued
L2-LinkConnect.request to L2. When L2 handover finished, L3 received
L2-LinkUp.indication from L2. It took 1ms for an L2 handover. Next,
MN sent the FNA message to NAR. Upon receiving the FNA message, NAR
started forwarding packets to NM. ICMP echo request packets were
sent at 10ms intervals. It was observed that zero or one ICMP echo
reply packet was lost during a handover.

Teraoka, et al. Experimental [Page 21]

RFC 5184 L2 Abstractions for L3-Driven Fast Handover May 2008

MN PAR NAR
| | |
|----- RtSolPr --->| |
|<---- PrRtAdv ----| |
| | |
+--- |------ FBU ------>| |
| | |------- HI ------>|
20ms| | | |
| | |<----- HAck ------|
| | | |
+--- |<-------------- FBack -------------->|
| | |
+-- disconnect | |
| 1ms| | |
| connect | |
8-10ms| | | |
| 7ms| | |
| | | |
| +----- FNA -------------------------->|
+-- |<------------------------ deliver packets
| | |

Figure 7: Measured Results

Appendix C. Example Mapping between L2 Primitives and the Primitives in
IEEE 802.11 and IEEE 802.16e

This section shows example mapping between the L2 primitives and the
primitives in IEEE 802.11 [7] and IEEE 802.16e [9] in Table 1.

Teraoka, et al. Experimental [Page 22]

RFC 5184 L2 Abstractions for L3-Driven Fast Handover May 2008

+-------------------+----------------------+------------------+
| L2 Primitive | IEEE802.11 | IEEE802.16e |
+-------------------+----------------------+------------------+
| L2-LinkStatus | PMD_RSSI | Available |
| | | |
| | PMD_RATE | |
| | | |
| L2-PoAList | MLME-SCAN | M_ScanScheduling |
| | | |
| | | M_Scanning |
| | | |
| L2-PoAFound | MLME-SCAN | M_Neighbor |
| | | |
| | | M_Scanning |
| | | |
| L2-PoALost | MLME-SCAN | M_Neighbor |
| | | |
| | | M_Scanning |
| | | |
| L2-LinkUp | MLME-ASSOCIATE | M_Registration |
| | | |
| | MLME-AUTHENTICATE | |
| | | |
| L2-LinkDown | MLME-DEASSOCIATE | M_Ranging |
| | | |
| | MLME-DISAUTHENTICATE | |
| | | |
| L2-StatusChanged | PMD_RSSI | M_Ranging |
| | | |
| | | M_ScanReport |
| | | |
| | | M_Scanning |
| | | |
| L2-LinkConnect | MLME-JOIN | M_MACHandover |
| | | |
| | MLME-ASSOCIATE | M_HOIND |
| | | |
| | MLME-REASSOCIATE | |
| | | |
| | MLME-AUTHENTICATE | |
| | | |
| L2-LinkDisconnect | MLME-DISASSOCIATE | M_Management |
| | | |
| | MLME-DEASSOCIATE | (Deregistration) |
+-------------------+----------------------+------------------+

Table 1: Mapping between L2 Primitives and the Primitives in
IEEE 802.11 and IEEE 802.16e

Teraoka, et al. Experimental [Page 23]

RFC 5184 L2 Abstractions for L3-Driven Fast Handover May 2008

Appendix D. Example Mapping of Primitives and IEEE 802.11

This section shows examples of the mapping between primitives and
IEEE 802.11 [7] parameters.

D.1. L2-LinkStatus

Most parameters of L2-LinkStatus are related to the configuration of
network-interface hardware. The operating system and the
device-driver module can easily collect such information. However,
to create the "Condition" parameter, the MN should maintain
statistics and parameters related to the current wireless
environment.

There are two sub-parameters of the "Condition" parameter: available
bandwidth and link quality level. The available bandwidth of the
current PoA can be obtained by maintaining the transmission rate
indication and the statistics of frame transmission every time a
frame is sent. A link quality level can be updated by maintaining
the following parameters and statistics every time a frame is
received: Received Signal Strength Indicator (RSSI), transmission/
reception rate indication, transmit/receive latency, bit-error rate,
frame-error rate, and noise level. Link quality level is divided
into five levels: EXCELLENT, GOOD, FAIR, BAD, and NONE, in descending
order. Some parameters can be managed by setting thresholds from
software. When the parameters cross the threshold, an interrupt is
generated for the software.

D.2. L2-PoAList

In IEEE 802.11 networks, L2-PoAList returns the detected APs whose
quality level exceeds the specified threshold for PoA candidates (by
the "PoA List" parameter). Therefore, an MN should always maintain
the database of available access points according to reception of
beacon frame, probe response frame, and all frames. This AP database
is managed in consideration of time, number of frames, and signal
strength. There are some kinds of network-interface hardware that
can notify events to operating system only when the desired event
occurs by setting a threshold from software. Moreover, MN can
transmit the probe request frame for access point discovery, if
needed.

D.3. L2-PoAFound

In IEEE 802.11 networks, L2-PoAFound is notified when new PoAs whose
link quality level exceeds the specified threshold are detected by
listening beacons or scanning APs. If the received frame (mainly the

Teraoka, et al. Experimental [Page 24]

RFC 5184 L2 Abstractions for L3-Driven Fast Handover May 2008

beacon or the probe response) is not in the AP database described in
Appendix D.2, then the link layer creates L2-PoAFound.indication.

For example, if the threshold of link quality level specified by
L2-PoAFound.request is GOOD, L2-PoAFound.indication is created and
sent to the upper layer when PoA's link quality becomes better than
GOOD.

D.4. L2-PoALost

In IEEE 802.11 networks, L2-PoALost is notified when an AP included
in the list of candidate APs is lost by listening beacons or scanning
APs. If the entry in the AP database described in Appendix D.2
expires, or link quality level goes under the threshold, then the
link layer creates L2-PoALost.indication. To calculate the link
quality level, the signal strength of the received frame (mainly the
beacon or the probe response) can be used. For example, if the
threshold of the link quality specified by L2-PoALost is BAD,
L2-PoALost.indication is created and sent to the upper layer when
PoA's link quality becomes worse than BAD.

D.5. L2-LinkUp

In IEEE 802.11 networks, L2-LinkUp is notified when association or
reassociation event occurs. When such an event occurs, MN receives
the association response frame or the reassociation response frame.
Immediately after receiving it, the link layer creates
L2-LinkUp.indication.

D.6. L2-LinkDown

In IEEE 802.11 networks, L2-LinkDown is notified when a
disassociation event occurs or when no beacon is received during a
certain time. When such an event occurs, MN sends the disassociation
frame to AP, or the entry of the specific AP is deleted from the AP
database described in Appendix D.2. By detecting such events, the
link layer creates an L2-LinkDown.indication.

D.7. L2-LinkStatusChanged

In IEEE 802.11 networks, L2-LinkStatusChanged is notified when the
radio signal strength of the connected AP drops below the specified
threshold.

Teraoka, et al. Experimental [Page 25]

RFC 5184 L2 Abstractions for L3-Driven Fast Handover May 2008

D.8. L2-LinkConnect

In IEEE 802.11 networks, each AP is identified by the BSSID and the
service set of several APs is identified by the SSID. When
L2-LinkConnect is used to connect a specific AP or a service set, the
link layer should set the Basic Service Set Identifier (BSSID) or the
Service Set Identifier (SSID). Also, the channel should be set
appropriately at the same time by searching the database described in
Appendix D.2. To connect to AP, MN should be authenticated by AP.
MN sends the authentication message to AP, and then MN sends the
association message to associate with AP.

D.9. L2-LinkDisconnect

In IEEE 802.11 networks, MN sends the disassociation message to AP
for disconnection. When L2-LinkDisconnect is used for disconnection
from the current AP, the link layer should send the disassociation
message to the appropriate AP, and stop data transmission.

Appendix E. Implementation and Evaluation of the Proposed Model

This section describes an implementation of the proposed link
indication architecture and its evaluation.

An IEEE 802.11a wireless LAN device driver was modified to provide
abstract link-layer information in the form of primitives defined in
Section 4. The modified driver has two AP lists. One contains the
device-dependent information such as RSSI, retransmission count,
various AP parameters, and various statistics. The device-dependent
information, except for the AP parameters, is updated whenever the
device receives a frame. If the received frame is the management
frame, the AP parameters are also updated according to the parameters
in the frame.

Another AP list contains the abstract information. The abstract
information is updated periodically by using the device-dependent
information. In the abstraction processing, for example, RSSI or the
retransmission count is converted to the common indicator "link
quality". In our outdoor testbed described below, the thresholds of
the RSSI value between the link quality levels were defined as
follows:

o EXCELLENT -- 34 -- GOOD

o GOOD -- 27 -- FAIR

o FIAR -- 22 -- BAD

Teraoka, et al. Experimental [Page 26]

RFC 5184 L2 Abstractions for L3-Driven Fast Handover May 2008

o BAD -- 15 -- NONE

L2-PoAList and L2-LinkStatus were implemented by using only the
abstract information. Thus, the upper layers can use information of
the current AP and the adjacent APs without depending on the devices.
L2-PoAFound or L2-PoALost is notified if the link quality rises or
falls beyond the thresholds when the abstract information is updated.
If the link quality of the current AP goes below the specific
threshold, L2-LinkStatusChanged is notified. L2-LinkUp or
L2-LinkDown is notified when an association with an AP is established
or destroyed. To realize L2-LinkConnect and L2-LinkDisconnect,
functions to connect or disconnect to the specified AP were
implemented. In these functions, since only abstract information is
needed to specify the AP, other layers need not know the
device-dependent information.

In our outdoor testbed, there are eight Access Routers (ARs) located
at 80-100m intervals. AP is collocated at AR. IEEE 802.11a was used
as the link layer. The same wireless channel was used at all APs.
Thus, there are eight wireless IPv6 subnets in the testbed. The
mobile node was in a car moving at a speed of 30-40 km/h. When link
quality of the current AP goes down, the mobile node executes
L3-driven handover, described in Appendix A. An application called
Digital Video Transport System (DVTS) was running on the mobile node
and sent digital video data at a data rate of about 15Mbps through
the wireless IPv6 subnet to the correspondent node, which replayed
received digital video data. In this environment, the L2 handover
required less than 1 msec, and the mobility protocol Location
Independent Networking in IPv6 (LIN6) [13] required a few msecs for
location registration. Thus, the total gap time due to the handover
was 3-4 msec. In most cases, there was no effect on the replayed
pictures due to handover.

Authors' Addresses

Fumio Teraoka
Faculty of Science and Technology, KEIO University
3-14-1 Hiyoshi, Kohoku-ku
Yokohama, Kanagawa 223-8522
Japan

Phone: +81-45-566-1425
EMail: [email protected]
URI: http://www.tera.ics.keio.ac.jp/

Teraoka, et al. Experimental [Page 27]

RFC 5184 L2 Abstractions for L3-Driven Fast Handover May 2008

Kazutaka Gogo
Graduate School of Science and Technology, KEIO University
3-14-1 Hiyoshi, Kohoku-ku
Yokohama, Kanagawa 223-8522
Japan

Phone: +81-45-566-1425
EMail: [email protected]
URI: http://www.tera.ics.keio.ac.jp/

Koshiro Mitsuya
Jun Murai Lab, Shonan Fujisawa Campus, KEIO University
5322 Endo
Fujisawa, Kanagawa 252-8520
Japan

Phone: +81-466-49-1100
EMail: [email protected]

Rie Shibui
Graduate School of Science and Technology, KEIO University
3-14-1 Hiyoshi, Kohoku-ku
Yokohama, Kanagawa 223-8522
Japan

Phone: +81-45-566-1425
EMail: [email protected]
URI: http://www.tera.ics.keio.ac.jp/

Koki Mitani
Graduate School of Science and Technology, KEIO University
3-14-1 Hiyoshi, Kohoku-ku
Yokohama, Kanagawa 223-8522
Japan

Phone: +81-45-566-1425
EMail: [email protected]
URI: http://www.tera.ics.keio.ac.jp/

Teraoka, et al. Experimental [Page 28]

RFC 5184 L2 Abstractions for L3-Driven Fast Handover May 2008

Full Copyright Statement

Copyright (C) The IETF Trust (2008).

This document is subject to the rights, licenses and restrictions
contained in BCP 78 and at http://www.rfc-editor.org/copyright.html,
and except as set forth therein, the authors retain all their rights.

This document and the information contained herein are provided on an
"AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS
OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY, THE IETF TRUST AND
THE INTERNET ENGINEERING TASK FORCE DISCLAIM 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.

Intellectual Property

The IETF takes no position regarding the validity or scope of any
Intellectual Property Rights or other rights that might be claimed to
pertain to the implementation or use of the technology described in
this document or the extent to which any license under such rights
might or might not be available; nor does it represent that it has
made any independent effort to identify any such rights. Information
on the procedures with respect to rights in RFC documents can be
found in BCP 78 and BCP 79.

Copies of IPR disclosures made to the IETF Secretariat and any
assurances of licenses to be made available, or the result of an
attempt made to obtain a general license or permission for the use of
such proprietary rights by implementers or users of this
specification can be obtained from the IETF on-line IPR repository at
http://www.ietf.org/ipr.

The IETF invites any interested party to bring to its attention any
copyrights, patents or patent applications, or other proprietary
rights that may cover technology that may be required to implement
this standard. Please address the information to the IETF at
[email protected].

Teraoka, et al. Experimental [Page 29]