Broadband Access networks
- a wide range of technology choices
loa.andersson@acreo.se
tanja.kauppinen@acreo.se
The diversified access
A broadband network - or any tele- or data communication network - logically breaks down in several discrete parts. On a high level we talk about home, access and core networks. The access network includes:
- the network termination point at the customer premises (sometimes only a jack)
- the cabling and protocols for transport of data up to the first active network equipment (sometimes called ”the first mile”)
- the network that aggregates the traffic to and from end-users onto the core network
For this part of the network there exists a plethora of solutions, developed to leverage existing infrastructure deployments and/or specific link technologies. The services and the networking technologies often have adapted to this diversification. The positive effect of this is that it is possible to deliver a wide range of services in the networks, the down-side is that it takes a lot of effort to adapt a new service to all access technologies.
Open networks
In the open network paradigm it is inherent that it would be beneficial to minimize the number of combinations of technologies used in the network, and particularly in access networks. The link technology that we have chosen to use for the Acreo test bed is Ethernet, the physical media is deliberately more diversified and we are going to perform some comparison tests between the different solutions.
The customer interface and the first mile traffic pattern
Acreo has spent some time on investigating what type of interface a representative user in the multi service open network will find optimal. For a household with more or less all of the communication services delivered over the converged network it is fairly easy to calculate what interface is needed and which is also economically feasible. Telephony, TV, Internet, HDTV and Video on Demand services will require at least 40-50Mbit/s for real-time sensitive traffic, a bit depending on how many channels you need to use at the same time. There is also a need for a substantial amount of best effort traffic.
The traffic pattern for household-based end-users has for a long time been said to be ”asymmetrical”, i.e. more traffic is sent towards this type of user than is generated by the user. This is e.g. the assumption that led to an asymmetric specification of the most used DSL technology - ADSL. Actual measurements in the test bed does not support this claim, it is rather that as soon as you have a symmetric access you tend to move towards a more symmetrical traffic pattern.
Based on the figures above a 100 Mbit/s Ethernet (Fast Ethernet, FE) interfaces is a fairly obvious choice: the cost is low, the availability is extremely good and it has the capacity needed. We currently don’t see a need for a higher bandwidth interface for household deployment, but in the not too distant future the costs for Gbit Ethernet (GE) interfaces will come down and be an alternative.
Business customers
Business customers and test activities focusing on their needs also have its place in the Acreo test bed. We are looking into different ways of creating secure corporate Virtual Private Networks (VPNs) carried across a public infrastructure. We are working on both L2 and L3 VPN solutions and specifications. This is also an area where Acreo is active in standards. A secure corporate network, a secure Intranet, has two aspects, traffic separation and data integrity. Traffic separation is handled by some type of tunnel technology, e.g. MPLS, and data integrity by end-to-end cryptography. We see tendencies that the interfaces for corporate VPNs to go beyond 100 Mbits, the architectural work we are doing will investigate what type of interface will be recommended for the corporate VPNs.
Access technologies
Currently we have the following last mile technologies in the test bed:
Fiber To The Home (FTTH) - a point to point solution on single mode and multi-mode fiber. The FTTH access is also tested in two radically different environments, a more rural area in Hudiksvall and urban areas in Stockholm and Hudiksvall.
Ethernet Passive Optical Networks (EPON) - a fiber based optical system for Ethernet access.
Asynchronous Digital Subscriber Loop 2+ (ADSL2+) - a copper based (using the last mile telephony copper network) access technology.
Wireless LAN (WLAN) - a technology to transport Ethernet over radio.
Fixed Wireless Access (FWA) – broadband wireless access technology for long reach coverage suitable in sparsely populated areas. Coverage radius up to 20km. There is no FWA installation in the test bed today but it is a possible future extension.
Category 5 (Cat5) - a twisted pair copper technology used e.g. in buildings, the bandwidth is distance dependent, but over 4-500 m it is possible to run e.g. Fast Ethernet.
It is likely that the number of access technologies will increase in the future as new technologies become available. It is also likely that the number of test users, i.e. end-users that use the Acreo test bed, will vary in the future. To test an access technology we will need a certain number of test pilots; it is however not in Acreo’s interest to expand the number of test pilots beyond the number needed for the tests.
Layer 2 and Layer 3
The discussion whether a Layer 2 (L2) or Layer 3 (L3) technology should be used in access networks has been going on since the Metro Networks were first built. The terms L2 and L3 refer to the Open Systems Interconnect (OSI) reference model, L3 specifies the network functions and L2 specifies the link functions needed.
It is important to remember that the decision of L2 and L3 is not a question of either or, but of how L2 and L3 is deployed and what roles they play. In any network, viewed end-to-end both L2 and L3 functions are used and needed. Also L2-networks of today tend to use some L3-functionality but do not implement a full L3 solution. However in a moderate size access network it is not strictly necessary to use L3 functionality, this could be handled by the service provider and the end-user. In the Acreo test bed we deliver services over access networks that are built according to both paradigms. So far we have not been able to demonstrate any significant performance or capability difference between the L2 and L3 access networks. The size of the access networks in the test bed is however limited and we will continue investigating and testing the scaling properties and the impact on scaling from different access technologies.
Access network topology
The test bed access network in Hudiksvall is build using a L2 manageable Gigabit Ethernet (GE) infrastructure connecting the L2 and L3 access switches. Each access switch connects up to 24 end-users into the access network.
The test bed in Stockholm has basically the same topology, though since there is only one group of end-users the network topology appear to be more meshed.
In both cases services are delivered form the service providers over the access network, by means of a high-end L3 router and L3 switches.
Results
The results so far show that Fast Ethernet in the first mile is a viable technology and that it can deliver the wide range of services needed in a household environment. More issues are found on the end-user premises than in the access network.