19-07-2014, 11:58 AM
Mobile Wireless Internet Radio Access Networks
Mobile Wireless.pdf (Size: 737.82 KB / Downloads: 73)
ABSTRACT
Cellular telephony networks depend on an extensive wired network to provide access
to the radio link. The wired network, called a radio access network, provides such
functions as power control and, in CDMA networks, combination of soft handoff legs
(also known as macrodiversity resolution) that require coordination between multiple
radio base stations and multiple mobile terminals. Existing RAN architectures for
cellular systems are based on a centralized radio network controller connected by
point-to-point links with the radio base transceiver stations. The existing architecture
is subject to a single point of failure if the RNC fails, and is difficult to expand
because adding an RNC is expensive. Also, although a network operator may have
multiple radio link protocols available, most RAN architectures treat each protocol
separately and require a separate RAN control protocol for each. In this article we
describe a new architecture, the OpenRAN architecture, based on a distributed
processing model with a routed IP network as the underlying transport fabric.
OpenRAN was developed by the Mobile Wireless Internet Forum IP in the RAN
working group. The OpenRAN architecture applies principles to the radio access
network that have been successful in reducing cost and increasing reliability in data
communications networks. The result is an architecture that can serve as the basis for
an integrated next-generation cellular radio access network
INTRODUCTION
Cellular telephony networks depend on an extensive wired network between
the core network and the radio transceivers that handle particular cells. This network,
called a radio access network (RAN), provides functions that coordinate access to the
radio link between multiple radio base stations and between mobile terminals. In this
article we discuss a new architecture for mobile wireless RANs. The architecture,
called the OpenRAN, is based on a distributed processing model with a routed IP
network as the underlying transport fabric. The next section briefly discusses existing
cellular RAN architectures and why a new architecture might be appropriate for
fourth-generation (4G) cellular systems. We then present the principal requirements
that drove the OpenRAN architectural design. In a later section the Open-RAN
architecture is summarized. The architecture consists of a collection of 77 atomic
functions grouped into 16 functional entities with 32 interfaces between them, so only
a brief overview of the architecture is possible to present here
2 Objectives
The long term vision of the OpenRAN architecture is to extend the peer-to-peer and
distributed Internet architecture to radio access networks, so a radio access network
becomes just another access network, like cable, DSL, Ethernet, etc.. The first step
toward this goal is to gather requirements. The second step is to determine the basic
functionality of a radio access network as a collection of functional entities in a
functional architecture, identifying the interfaces between the functional entities. The
third step is to describe which interfaces are open and how they can be implemented
using IETF protocols, or protocols based on IP but designed specifically for the radio
access network functions. The fourth step is to trace back the architecture to validate
that it does, in fact, meet the requirements. The objectives of the first version of this
report are to address steps one and two in the above process; future versions of this
report will address the remaining steps.
OpenRAN ARCHITECTURE
The OpenRAN architecture was developed by partitioning a RAN into atomic
functions, then grouping the atomic functions into functional entities based on the
requirements. Interfaces between the functional entities were then identified, and the
nature of the traffic over those interfaces was characterized as a first step toward
possible protocol development on those interfaces that are declared open. The
following subsections describe the functional entities, separated into control and
bearer planes. The radio layer 1 functions, and the operations , administration , and
management functions were kept separate because they span control and bearer.
Transport functional entities were also developed, but were not included in the
architecture because they are described more completely in a separate report devoted
exclusively to the problem of using IP transport in RANs[4]. Figure 2 contains an
illustration of the architecture, including numbered interfaces between the functional
entities.
FUTURE SCOPE
The focus in version 1 of the OpenRAN architecture has been on identifying the
radio-specific atomic functions of the RAN and grouping them into functional entities
that can be implemented as a distributed system. The working group is currently
working on mechanisms for pushing mobility management, QoS, and security/AAA
from the radio network layer into the transport layer, where they would be common
across all supported radio link protocols.A common radio network layer protocol for
multiple radio link protocols was identified as another area for future work.. Having a
common radio network layer protocol for controlling multiple radio link protocols
facilitates pushing mobility management, QoS, and security/AAA to the transport
layer, and in addition simplifies inter-radio link protocol handoff, allowing network
operators to optimize use of spectrum assets. However, there are questions about how
broad such a protocol can be, given the diversity
CONCLUSION
The OpenRAN architecture presented in this article is a first step toward an all-IP
radio access network. The architecture as it currently stands describes how to
decompose radio access network functionality in a way that allows a distributed
implementation and opens the door to implementing functions common among
multiple radio link protocols, namely mobility management, QoS, and security/AAA,
at the transport layer instead of in separate radio network layer protocols. There is still
much work to be done on key issues involving how to implement common functions
and on a common radio network layer protocol, and on interoperable network
management.