Content uploaded by Henrik Abramowicz
Author content
All content in this area was uploaded by Henrik Abramowicz on Jun 18, 2013
Content may be subject to copyright.
Ambient Networks
–
An Architecture for Communication Networks Beyond
3G
Norbert Niebert
1
, Andreas Schieder
1
, Henrik Abramowicz
1
, Christian Prehofer
2
,
Holger Karl
3
1
Ericsson Research, Corporate Unit
{Norbert.Niebert, Andreas.Schieder, Henrik.Abramowicz}@ericsson.com
2
DoCoMo Communications Laboratories Europe, prehofer@docomolab-euro.com
3
Telecommunication Networks Group, Technical University of Berlin,
hkarl@ieee.org
1 Introduction
In today’s wired and wireless networks, the trends in networking technology very much
point to a dominance of Internet technology with all its flavours. IP is the common
“lingua franca” to enable the exchange of data across various networks. There is,
however, an increasing divergence in the network control layer: different control
environments are established to facilitate services like VPN’s, security, mobility, QoS,
NAT, multicast, etc. For a multitude of services, data might still be handled by uniform
Internet networking, but the control of such services is becoming increasingly
fragmented. More and more, the network as a whole therefore diverts from the pure end-
to-end view of the Internet. Furthermore, applications would like to rely on enhanced and
consistent support from the network for the complete delivery chain.
This lack of a common control layer for joining the services – in a wide sense of the word
– of multiple networks represents a crucial challenge both technically and from a user
perspective. Usage scenarios that should be realizable in the mid-term future include
utilization of multiple devices, multiple networks, and multiple access technologies in an
integrated fashion. This is not easily controllable or manageable with today’s
technologies. These usage scenarios partly evolve from the vision of “Ambient
Intelligence”, developed and popularized by various European research efforts. The core
notion of this vision is to make technology invisible and useable, whenever necessary.
This notion holds in particular for networking technologies – networking should be
available whenever required in the most appropriate and affordable form.
Our approach, which we call Ambient Network, aims to provide a domain-structured,
edge-to-edge view for the network control. In this way, an Ambient Network is expected
to embrace the heterogeneity arising from the different network control technologies such
that it appears homogeneous to the potential users of the network services. The vision is
to allow the agreement for co-operation between networks on demand, transparent and
without the need of pre-configuration or offline negotiation between network operators.
End-users are increasingly not just owners of a terminal or a PC, they own and operate
effectively a network of devices in their homes, offices and around the body.
Consequently, they are included in this network of co-operation, treating them as
operators of special, low-complexity networks. This approach generalizes to many
different kinds of networks that are currently appearing, such as inter-vehicle networks,
body area networks, or sensor networks: By making every device a network, the network
is the primitive building block of our architecture, allowing all types of networks to be
integrated into a larger system.
This concept of Ambient Networking is guided by market and business analyses carried
out in European research projects as well as by the conclusions and technology concepts
discussed within the WWRF (Wireless World Research Forum [2]). WWRF gathers
research efforts targeting mobile networks beyond 3G. The forum provides an excellent
opportunity to understand the issue currently discussed in the context of networks beyond
3G and the relations between these issues (see Books of Vision [3] and [4]).
In the following sections, we briefly describe existing approaches for next generation
wireless networks. Based on this description, we discuss in Section 3 how we imagine to
progress from simple All-IP networks to Ambient Networks. Section 4 investigates how
Ambient Networks can be built, with some details highlighted in Section 5. Section 6
concludes the paper.
2 Previous efforts
To put these goals and requirements into perspective, let us have a brief look at two
related research efforts.
2.1 Architecture discussion in the Internet community
Research in the Internet community on future network architectures is mainly influenced
by the discovered deficiencies of the current Internet, where mobile networks and
mobility aspects are treated with comparatively low priority. The motivation for
reconsidering the design of the Internet is its current widespread and its impact on social,
economic and political aspects [6], [7]. Efforts are made to define a common set of
architectural principles and tenets, which guide the development of a new Internet
architecture. In parallel to these efforts, which adopt a top-down analysis, several
technology-focused activities exist working on solutions for new naming and addressing
schemes, QoS support and routing (e.g. [5]).
Interestingly, one of the findings made within these research activities is the concept of
separating the Internet into domains or realms, which are on a higher logical level
interconnected to build the Internet. The separate domains are defined to encompass
networks deploying different technologies than the ones used in the existing Internet.
This approach is supposed to ease technological advances, as the localized introduction
would be made possible. Still, the definition of the required border functions, translating
between different domains, has not been provided yet, except for some conceptual
approaches such as described in reference [8].
A second approach of the Internet research community is the work on overlay networks
e.g. introduced to improve network reliability or to derive a network structure
implementing different routing paradigms. In reference [9] such an overlay network is
described to implement an architecture tailored to the demands of content delivery, which
imposes modifications on the addressing and routing schemes implemented.
While all of these research efforts are essential to build upon, we believe that the
emerging needs of future wireless and mobile networks are not fully satisfied. Issues
concerning the (potentially automated) control of larger networks are not conclusively
addressed, how to combine smaller networks into larger ones, how to support mobility, or
heterogeneity are other exemplary shortcomings. The following sections will analyze
these needs and the solutions proposed by Ambient Networking.
2.2 IP2 Network Architecture initiative
The migration of mobile networks to an IP based network architecture is strongly
favoured by mayor players in the market (see e.g. [1]). NTT DoCoMo has for example
described an IP based IMT network platform (IP
2
) as successor of the existing 3G mobile
networks already in 2001. The concept has been introduced to meet the requirements
expected to arise when aiming at a mass-market support for mobile multimedia services.
The key requirements identified by IP
2
are:
• Support of multimedia traffic which equals 80% of the overall load in the cellular
system,
• Efficient support for mobility,
•
Support for diversified radio accesses,
• Support for a seamless service provisioning, and
• Open interfaces to allow external application providers to use the IP
2
communication
infrastructure.
The approach taken within IP
2
is mainly characterized by two design choices. First, it is
envisaged to provide an IP based transport network, which provides access to different
radio access networks. These networks can deploy various link level technologies, but are
expected to provide an IP based interface to allow connecting them to the IP
2
transport
network. Second, connectivity and control functions are separated and all control
functions are logically grouped in two domains referred to as Network Control Platform
(NCPF) and Service Support Platform (SSPF).
3 From All-IP networks to Ambient Networks
3.1 Current view on All-IP
Ambient Networks are based on all-IP based mobile networks and can be regarded as the
outcome of a continued adoption of Internet design principles. All-IP based mobile
networks can be characterized by – among other aspects - a clear separation between
transport and control related tasks. The functions concerned with either of these tasks are
grouped in two distinct layers to ease the independent development in both areas.
Additionally, the transport layer is supposed to be IP based, i.e. IP packets are the
smallest denominator of all transport layers.
The Ambient Network concept adapts these tenets and assumes the presence of a layer to
ensure basic connectivity between different networks, constituting diverse addressing
domains, participating in the creation of the Ambient Networks federation. This layer will
remain based on the Internet Protocol. The control functions already identified and
isolated by All-IP networks are extended to the “Ambient Control Space” (see Section
4.2), which embraces a well defined set of control functions required to guarantee the co-
operation between networks. The control space is thus no longer only concerned with the
management of the transport functions within its own network domain, but is additionally
responsible for the establishment of agreements with neighbouring networks to provide
end-to-end services through the setup of a service chain involving several networks.
3.2 How Ambient Networks extend All-IP networks
Ambient Networks extend All-IP networks by several innovations. The extensions build
on the demand for enabling communication between different social and economical
realms as identified by the Internet research community. The three main innovations are
network composition (beyond simple internetworking), enhanced mobility and effective
support for heterogeneity in networks.
Network Composition: One basic mechanism of our approach is the dynamic, instant
composition of networks (Figure 1 shows an example). Composability as required for
ambient networking goes beyond what the Internet and mobile networks provide today.
Internetworking shall happen not only at the level of basic addressing and routing, but
additional functions for incorporating higher layer support (such as content distribution or
service control functions) are required. Ambient Networks will deploy a universal
framework for network composition, as an approach for building a unified
communication support out of the resources of individual networks, which can be
specialised for particular types of access technologies or business models. Network
composition in Ambient Networks has to function across operator and technology
boundaries, provide a security framework and has to be executable without user
involvement. In addition, the execution of the composition process has to be rapid in
order to follow fast topology changes as expected for example for mobile Personal Area
Networks (PAN).
Another example where instant network composition is relevant is the joining of operator
networks, which today are based on explicit, human-negotiated and –executed
agreements and are therefore too slow and cumbersome to setup for, e.g., rapid service
roaming. Rather, access, interconnectivity, and service level will be associated on the fly
between two networks.
Figure 1: Ambient Networks enables the concept of instant composition of networks belonging to
different business entities
Mobility: Existing IP-based mobility solutions target either intra-domain mobility within
a static network architecture or roaming solutions across domain boundaries. Ambient
Networks focus on integrated mobility concepts not only applicable for both of the above
scenarios, but also integrating localised communications, e.g. in PANs, and device-device
interactions. In dynamically composed network architectures, mobility of user group
clusters can support effective local communication. Furthermore, mobility must interact
efficiently with the control interfaces needed to enable quality of service and optimal
routing and re-routing of individual multimedia flows. An Ambient Networks mobility
solution will have to work well across business and administrative boundaries, which
requires solutions for the security issues of inter-domain operation.
Heterogeneity: Ambient Networks will be based on a federation of multiple networks of
different operators and technologies. On the one hands, this leads to increased
affordability of ubiquitous communication, as the user has full freedom to select
technology and service offering and the investment needs for new networks are reduced.
On the other hand, networks will have to integrate the capabilities of different
technologies to an end-to-end, seamless and secure solution for the user.
Ambient Networks takes a new approach to embrace heterogeneity visible on different
levels, such as link technologies, IP versions, media formats and user contexts. Diversity
of access links, especially of links provided by mobile networks, is supported by a
generic link layer concept, which will efficiently enable the use of multiple existing and
new air interfaces (see section 5.2). Ambient Networks also consider the implications of
heterogeneous wireless systems on the overall network, especially the impact on end-to-
end QoS and multimedia delivery. In particular, the novel concept of network
composition will include the negotiation between different networks regarding their
capabilities, e.g. regarding quality of service. Ambient Networks provide an integrated
framework for enhanced support of multimedia delivery in heterogeneous environments
by embedding novel media flow routing and transport functionalities into the overall
Ambient Network architecture (see section 5.1).
4 How to build Ambient Networks
The vision of ambient networks is that users and operators can jointly exploit available
radio and network resources for a broad range of services. To enable this vision, the
concept builds on a common set of control functions, dynamically (re-) configurable and
universally available.
4.1 Design principles for Ambient Networks
To evolve today’s mobile networking into this vision, the current systems need to be
analysed, especially from a user’s perspective as outlined above. Currently, the lack of
commonly available and configurable control functions and dependence on network
technologies and ownership is a major obstacle to further rapid growth. Ambient
Networks takes up the challenge of defining an essential set of universally available and
usable control functions. To achieve this, it defines a conceptual framework including the
control functions necessary to achieve the required network capabilities. This framework
is based on the following three principles.
Principle #1: Ambient Networks build upon open connectivity and open
networking functions
One basically new way of defining networking in Ambient Networks is to remove
architectural restrictions on who or what can connect to what. Compared to existing inter-
networking, the goal is to enable all networking services for connected networks instead
of connected nodes, e.g. quality of service or media delivery support. This is motivated
by the observation that current, node-centric designs fail for many scenarios including
PANs, moving networks or sensor networks, when connecting such networks to other
networks. In general, we will always assume a network at the end of the communication
flow. Hence, we talk about “end-environments” rather than end nodes.
The Ambient Networks concept defines a set of support functions required to satisfy the
business needs of the operators of such end-environments, where operators can be
commercial entities as well as end-users. For such an end-environment, capability
offerings are broken down to their nucleus, which is the end-to-end relationship between
control functions. The challenge is to provide suitable mechanisms to enable any such
relationship regardless of whether the partners reside in an operator’s network or an end-
user’s terminal.
Principle #2: Ambient Networks are based on self-composition and self-
management
While composing networks is an easy task when only packet forwarding is concerned,
advanced end-to-end functions like QoS and security as well as mobility are currently
very difficult to establish across network boundaries. Ambient Networks treat network
composition and reconfiguration in a self-managed way as a guiding design principle for
the research work. The goal is to use network composition and self-management as basic,
locally founded building blocks of a networking architecture. This includes self-
composition across different business domains. These features will broaden the business
case for the operator and enable a fast introduction of new services in all connected
networks.
Principle #3: Ambient Networks functions can be added to existing networks
In today’s networks, there is a large degree of homogeneity on the very basic
connectivity/packet forwarding functions, but the network control is distributed over
multiple layers and specific to network technology, operator and even implementation.
As a starting point for Ambient Networks, the connectivity and control level are logically
separated. The control level, referred to as the Ambient Control Space (see next section),
can enhance existing technologies with distinct control functions, which are compatible
across all domains of an Ambient Network.
An exemplary networking scenario of a composed ambient network is shown in Figure 2.
With the above design principles, Ambient Networks will provide end-to-end services for
such a composition of heterogeneous networks.
End-to-end, mobility aware Connectivity
(QoS, media delivery,…)
AN 0
mobile
node
AN 1
e.g. PAN
AN n-1
Access
Netw.
AN n
Core
Netw.
Figure 2: Composed Ambient Network
This approach yields the following advantages:
Network operators can decide what level of support they want to give to users and
business partners, based on flexible sets of network control and composition
functions. Affordable and simple as well as service rich networks can be built without
restricting the integration into the global connectivity. Existing networks can be
integrated and furthermore new technologies can be introduced in a coherent way.
End-users select between different control functions on their devices or personal
networks, which provides the feature rich service environment they like to use.
Innovation is nurtured by the opportunity to develop any service on top of the basic
connectivity with as few barriers as possible because of the simple but extendable
connectivity interfaces.
Service Providers can easily address large numbers of users and deliver their specific
services to them without having to worry about their users’ access or network
capabilities.
4.2 Ambient Control Space concept
Following the three principles discussed above, the notion of the Ambient Control
Space is introduced to encompass all control functions in a certain network domain. The
Ambient Control Space together with a (possibly legacy) connectivity network is called
an Ambient Network.
The Ambient Control Space (see Figure 3) hosts a set of control functions, which might
be missing in certain instantiations of the control space. Examples for such functions
include the support for mobility or multi-access networks, but also more abstract
functions like the provisioning of context information.
The main characteristics of an Ambient Network are:
• It provides well defined control interfaces to other Ambient Networks and to service
platforms or applications;
• It provides (at least a subset of) the Ambient Control Space functions;
•
It can be dynamically composed of several other Ambient Networks to form a new
Ambient Network.
Cooperating Ambient Networks could potentially belong to separate administrative or
economic entities. Hence, Ambient Networks provide network services in a co-operative
as well as competitive way. When Ambient Networks and their control functions are
composed, care must be taken that each individual function controls the same resources
as before: by composing two Ambient Networks, resources shall not become a common
good but rather a good that can be traded.
4.3 Ambient Network Interfaces
Control spaces spanning several domains are required, as users will move among them
and want to exploit the capabilities of as many domains as possible; they will assume
their personal services to be at their disposal regardless of their location and physical
connections utilised. The intrinsic complexity of domains might vary and economic
considerations will decide the feasibility of different approaches. To allow cooperation
across different Ambient Networks, the Ambient Network Interface (ANI) is provided.
It offers standardised means to connect the functions of an ambient control space with
functions of another domain (see Figure 3). It also advertises the presence of control
space functions in adjacent domains to allow for a dynamic set of control functions in
each.
Accessing the services of an Ambient Network happens via the
Ambient Service
Interface (ASI). Even in a composed Ambient Network, only a single homogeneous
control space is visible to external entities (Figure 3). An application or service, making
use of the functions in the control space, will always find the same environment
regardless of which Ambient Service Interface it is currently connected to. The
mechanisms implemented in the ambient control space and the Ambient Network
Interface will ensure this.
Domain
Management
Ambient
Network
Interface
Mobility
Multiaccess
Ambient
Service
Interface
Media
Delivery
Context
P rovisioning
Security
Ambient
Connectivity
Connectivity
Controller
Ambient
Network
Interface
Ambient
Network Interface
Ambient
Service Interface
Domain
Management
Mobility
Multiaccess
Media
Delivery
Context
Provisioning
Security
Ambient
Connectivity
Connectivity
Controller
Domain
Management
Mobility
Multiaccess
Media
Delivery
Context
Provisioning
Security
Ambient
Connectivity
Connectivity
Controller
Network A
Network B
Network
A+B
Domain
Management
Ambient
Network
Interface
Mobility
Multiaccess
Ambient
Service
Interface
Media
Delivery
Context
P rovisioning
Security
Ambient
Connectivity
Connectivity
Controller
Ambient
Network
Interface
Domain
Management
Ambient
Network
Interface
Mobility
Multiaccess
Ambient
Service
Interface
Media
Delivery
Context
P rovisioning
Security
Ambient
Connectivity
Connectivity
Controller
Ambient
Network
Interface
Domain
Management
Ambient
Network
Interface
Mobility
Multiaccess
Ambient
Service
Interface
Media
Delivery
Context
P rovisioning
Security
Ambient
Connectivity
Connectivity
Controller
Ambient
Network
Interface
Ambient
Network Interface
Ambient
Service Interface
Domain
Management
Mobility
Multiaccess
Media
Delivery
Context
Provisioning
Security
Ambient
Connectivity
Connectivity
Controller
Domain
Management
Mobility
Multiaccess
Media
Delivery
Context
Provisioning
Security
Ambient
Connectivity
Connectivity
Controller
Network A
Network B
Network
A+B
Domain
Management
Ambient
Network
Interface
Mobility
Multiaccess
Ambient
Service
Interface
Media
Delivery
Context
P rovisioning
Security
Ambient
Connectivity
Connectivity
Controller
Ambient
Network
Interface
Domain
Management
Ambient
Network
Interface
Mobility
Multiaccess
Ambient
Service
Interface
Media
Delivery
Context
P rovisioning
Security
Ambient
Connectivity
Connectivity
Controller
Ambient
Network
Interface
Figure 3: Co-operating Ambient Control Spaces appear as a single domain to external service users
A particular issue is to ensure scalability of these interfaces to be applicable for small
personal networks up to large-scale networks; a related issue is the universality of the
approach covering all types of networks.
In summary, the ANI and ASI have the following properties:
• They offer similar functions regardless of the nature of the Ambient Network,
•
support a simple plug&play connection between Ambient Networks,
• ensure network reconfigurability,
•
provide a single interface to the outside and
• are designed to support moving networks.
5 Ambient Networks in detail
In this section two functions envisaged for Ambient Networks are presented in more
detail. They have been selected as they present functions logically located at the upper
and lower boundaries of the Ambient Networks scope. Several other key functions, such
as quality of service, self-management, context management and security are not detailed
in this paper.
5.1 Ambient Media Delivery Overlay
Media distribution and delivery is foreseen to become an important task within Ambient
Networks. Multimedia services have different requirements when it comes to required
media delivery support functions (e.g. conversational multimedia, download, messaging,
streaming, multicast/broadcast). Synergies can only be expected from providing a
common media delivery support layer to all kind of multimedia services. However, this
media delivery layer needs to fulfil all requirements posed by existing multimedia
services and it needs to be extendable in order to cope with future service requirements.
A particular challenge is the support of “ambient” multimedia services, integrating
different end-devices and various access technologies in a seamless way for the sake of
an increased service performance and end-user convenience.
In Ambient Networks the media delivery support layer described above is realized as an
overlay network.
The Ambient Network logical nodes constituting the Media Delivery Overlay Network
are referred to as “MediaPorts”. MediaPorts could be co-located with or residing at
physical nodes of the interconnectivity layer. The media delivery support functions
provided by the Ambient Media Delivery Network are exposed to the service layer via a
well-defined interface (see Figure 4).
Basic support functions provided by the media delivery overlay network are for instance
media routing, caching and adaptation.
Media adaptation refers to capabilities for adapting media content to different data rates
(rate adaptation) and for converting between different media formats (transcoding). A
MediaPort can provide the required adaptation functions by itself or it manages and
controls the required adaptation resources. The needs of commercial content providers
need to be considered carefully, since they often would like to avoid uncontrolled
transformation of their content.
MediaPorts can also provide caching functionality, which refers to the capability to store
media or certain fractions of media in order to avoid long haul communication paths
between media source and media clients (sink). Media caches will primarily be used for
content for which a high demand is expected or has already been monitored.
Service Layer
DRM, session control, user interaction
Fiber
…
…
Global
Transwitching
Backbones
…
Regional
Backbones
Cable
3G
2G
WLAN
ADSL
Access
Networks
Mall
Home
IntraNet
Vehicular
PAN
Private
Networks
Telia, Tele2, … Skanova, … Sprint, WorldCom
Service layer API
MediaPorts
Ambient Multimedia
Delivery Network
Transport
layer API
Service LayerService Layer
DRM, session control, user interaction
Fiber
…
…
Global
Transwitching
Backbones
…
Regional
Backbones
Cable
3G
2G
WLAN
ADSL
Access
Networks
Mall
Home
IntraNet
Vehicular
PAN
Private
Networks
Telia, Tele2, … Skanova, … Sprint, WorldCom
Service layer API
MediaPortsMediaPorts
Ambient Multimedia
Delivery Network
Transport
layer API
Figure 4: The Ambient Media Delivery Network is realized as an overlay network above the IP
transport layer
Apart from providing media adaptation and caching capabilities, MediaPorts also act as
routers for media data within the overlay network. New routing protocols within the
overlay network are required to optimise for media delivery, in particular for real-time
services. For instance, a couple of researchers have started to investigate gains from
multi-path diversity when delivering real-time content across networks with redundant
paths (see e.g. [12]). The idea is to utilize more than just one path through the network
between a sender and a receiver in order to increase the error-robustness of real-time
services. Although the usefulness of path-diversity could be proved, there is currently
little discussion on how path diversity could be supported for real-time services in a
generic way. Note that support for multi-path routing is not only useful for increased
service robustness, but also for multi-access scenarios where consumers like to interact
with a multimedia application via various devices and different access technologies.
Another area for research is the investigation of new metrics upon which routing
decisions should be made. Those metrics could for instance take into account
characteristics of the media content, link characteristics and user preferences for
delivering media content in the best possible and most efficient way. Utilizing
information from lower transport layers within the Ambient Media Delivery Network, the
routing and adaptation decisions can be further optimised
5.2 The Generic Link Layer (GLL) for Multi-Radio Access
For today’s mobile communication systems, where specific mobile networks use
particular radio access technologies, the approach of highly optimized radio protocol
layers has proven to be an efficient design choice. However, the property of
heterogeneity of future Ambient Networks means that connectivity to mobile terminals
will be provided via a multitude of radio access technologies. This will enable dynamic
selection and configuration of different radio access technologies to allow cost-effective
data transmission based on service requirements, link quality and radio resource
consumption. To account for this new networking paradigm, it is required to provide
methods for cooperation of different radio access technologies and networks. Ambient
Networks will provide two such approaches. First, multi-radio resource management
functions allow coordinated allocation of heterogeneous types of radio resources. This
will enable load balancing across different radio access technologies and increase the
wireless system capacity. Second, a generic link layer will perform seamless and efficient
radio protocol reconfiguration for dynamic changes of radio access technologies. The
generic link layer will exploit commonalities of radio access technologies and allow
cooperation of radio access networks on the link layer. The generic link layer provides a
toolbox of configurable link layer processing functions, in order to achieve universal link
layer processing for a multitude of radio access technologies. For example, the generic
functions comprise compression of higher layer protocol headers, segmentation of higher
layer data, recovery of transmission errors by means of automatic repeat request (ARQ)
and data scheduling. A GLL configuration for a particular radio access technology
includes the selection of the appropriated link layer functions from the toolbox of
functions, as well as determining suitable protocol parameters and timers. The
configuration of the generic link layer toolbox has to be controlled by a configuration
management function in case parameters cannot be determined autonomously by e.g.
measurements of link round-trip times. Still, the generic link layer will also allow the
inclusion of radio specific functions for such cases in which a generalization is
impossible or inefficient. An example of a radio specific function is a medium access
scheme designed for the particular properties of a certain physical layer technology. A
reconfiguration of the GLL due to a change of radio access technology will be seamless,
without a loss of link layer state or buffered data. It is therefore required that the GLL
reconfiguration is coordinated with mobility management functions and context transfer
at inter-system handover is supported. As a result, the generic link layer provides
compatible radio link layers for different radio access technologies, which allow to hand
over link state information. This enables seamless data transmission even across access
network boundaries (see Figure 5). A more concise overview of the generic link layer can
be found in [13].
Multiple link layers
IP network
L2 B
IP
L1 B
L2 A
L1 A
Generic link layer
IP
L1 B
L1 A
GLL
a) b)
RAN A RAN B
LL2
IP
L2 A
IP
L1 A
LL1
IP
L2 B
IP
L1 B
IP network
RAN A RAN B
LL2
IP
GLL
IP
L1 A
LL1
IP
GLL
IP
L1 B
Figure 5: Multiple link layer scenario compared to generic link layer scenario
6 Conclusions
We have presented a new networking concept referred to as Ambient Networks, which
aims at enabling the co-operation of heterogeneous networks belonging to different
operator domains. It can be regarded as extension and continuation of the network
development starting from the All-IP networking vision and the current trend in the
Internet community to structure the Internet into smaller domains. It will enable the
introduction of new networking technologies as well as new services in a flexible way.
Co-operation between networks plays a key role in our concept, as communication end-
points are treated as networks as well. Additionally, we demand the possibility for instant
network composition to allow rapid adaptation of the network topology as required for
moving networks. The Ambient Control Space turning today’s networks into Ambient
Networks will enable this.
The motivation for our concept has been discussed and its main features were introduced.
Two Ambient Control Space functions “Media Delivery” and “Generic Link Layer” were
presented in more detail.
The issues presented in this paper are topics for research in the Ambient Networks
project, which is part of the EU’s 6
th
Framework Programme. In total 37 organizations
from Europe, Canada, Australia and Japan are involved in this Integrated Project, which
will run from 2004-2005 in its first phase.
7 Acknowledgements
The authors would like to thank Bengt Ahlgren, Robert Hancock, Uwe Horn, Göran
Malmgren, Toon Norp, Frank Reichert, Joachim Sachs and Jens Zander whose
contributions substantially shaped the vision of Ambient Networks.
8 References
[1] H. Yumiba, K. Imai, and M. Yabusaki, “IP-Based IMT Network Platform”, IEEE
Personal Communications Magazine, vol. 8, no. 5, pp.18-23, October 2001
[2] Website, “The Wireless World Research Forum”, http://www.wirless-world-
research.org
[3] Wireless Strategic Initiative, “Book of Visions 2000”, http://www.wireless-world-
research.org/Bookofvisions/BookofVisions2000.pdf
[4] Wireless Strategic Initiative, “Book of Visions 2001”, http://www.wireless-world-
research.org/BoV1.0/BoV/BoV2001v1.1B.pdf
[5] Ion Stoica, Daniel Adkins, Shelley Zhuang, Scott Shenker, and Sonesh Surana,
"Internet Indirection Infrastructure", Proc. ACM Sigcomm 2002, Pittsburgh, PA,
USA, August 2002
[6] D. D. Clark, J. Wroclawski, K. Sollins, and R. Braden, “Tussle in Cyberspace:
Defining Tomorrow’s Internet”, Proc. ACM Sigcomm 2002, Pittsburgh, PA, USA,
August 2002
[7] Clark, D., Sollins, K., Wroclawski, J., and Faber, T., "Addressing Reality: An
Architectural Response to Real-World Demands on the Evolving Internet." Proc.
ACM Sigcomm FDNA 2003 Workshop, Karlsruhe, August 2003.
[8] Jon Crowcroft, Steven Hand, Richard Mortier, Timothy Roscoe, and Andrew
Warfield "Plutarch: An Argument for Network Pluralism", Proc. ACM Sigcomm
FDNA 2003 Workshop, Karlsruhe, August 2003.
[9] Gritter, M., and Cheriton, D. R., “An Architecture for Content Routing Support in
the Internet”, In proceedings of 3
rd
USENIX Symposium on Internet technologies
and Systems (USITS’01), March 2001, pp. 37-48
[10] Dinesh C. Verma, Content Distribution Networks, Wiley 2002, ISBN 0-471-44341-7
[11] K. Obraczka, Multicast transport protocols: a survey and taxonomy. IEEE
Communications Magazine 36(1): 94-102, Jan 1998.
[12] Shiwen Mao, Shunan Lin, Dennis Bushmitch, Sathya Narayanan, Shivendra S.
Panwar, Yao Wang, and Rauf Izmailov, Real Time Transport With Path Diversity,
The 2nd Annual New York Metro Area Networking Workshop, Sep 2002, New York
[13] Joachim Sachs, “A Generic Link Layer for Future Generation Wireless Networking”,
Proc. IEEE International Conference on Communications (ICC) 2003, Anchorage,
AK, USA, May 2003.
