Local loop unbundling (LLU) has long been the main enabler of broadband competition in the DSL era by allowing third parties to install their own DSLAM termination equipment in central offices (COs) and lease the access copper from incumbents. But it is becoming redundant or at least uneconomic in the age of super broadband as fiber pushes deeper and is combined with advanced vectoring technologies that entail deployment of DSLAMs close to the home in street cabinets. It becomes uneconomic or impractical to deploy multiple DSLAMs so that each competitive operator has one to serve its own customers, not least because vectoring yields the best performance when all lines are terminated at a single DSLAM. This has the effect of reintroducing a form of monopoly at the infrastructure level as competitors no longer have their own DSLAM capacity.
To address this situation an alternative approach was been developed called Virtual Unbundled Loop Access (VULA), a term we first saw used by UK regulator Ofcom in a 2010 report on future competition in the super broadband era. VULA virtualizes the access network including the DSLAMs, with capacity then sold to competitive operators at the Ethernet bitstream level rather than the higher IP layer. This enables the operators to offer some competitive features such as class of service differentiation, but not as many as they can with traditional physical unbundling. For example they may depend on the incumbent operator for IP multicast support and therefore their ability to offer IPTV services. In some case such multicast support was not provided, as is the case in Spain with the wholesale VULA offering from Telefonica.
Just about all European countries now offer some form of VULA but with significant variations between them, so that while a few like the UK, Italy and Belgium allow interconnection with competitive networks at the exchange level, others like Spain, Switzerland, the Netherlands and Turkey do not.
The greater issue lies in the level of control competitive operators have over their services, which is inferior to traditional LLU in the case of many wholesale VULA offerings. Given that the incumbent will effectively be providing the infrastructure as fiber gets closer to the home, the only recourse is a full blown virtualized environment that allows competitive operators total control at the network level. This argument is cogently made in a white paper just published by Assia, a provider of broadband optimization technologies including vectoring. The company argues that Software Defined Networking (SDN), currently being deployed across some enterprise networks and cloud services, holds the keys to successful virtualization of the access network. We should note that Assia is launching its SDN for access networks at the Broadband World Forum 2014 later this month in Amsterdam, so does have an axe to grind. But the paper promoting its new offering it calls Software Defined Access Networking (SDAN) makes a strong case for applying SDN principles to the access layer.
Definitions of SDN vary, but in this context we mean simply separating the underlying transmission of data including IP packet forwarding across the access network from higher level traffic management and separation between services. SDN, or in this case SDAN, provides open interfaces through which network traffic flow can be controlled in software operating above the underlying packet forwarding. Assia contends that SDAN can then reproduce the operational and functional aspects of LLU almost exactly, while benefiting from the greater flexibility and scale economies of virtualization. With SDAN it is no longer necessary to dedicate space in the CO or street cabinet for separate equipment, while the access fiber network out as far as street cabinets where the DSLAMs now resides, or even to the home in FTTH deployments, can now be shared.
A key point is that with VULA as currently deployed across Europe, the incumbent or wholesale provider takes care of nearly all control functions, such as traffic management through tunnels or VLANs in the backhaul network, configuring the Access Node where the fiber meets the final cooper loops to the home, performing line diagnostics, and operations such as troubleshooting. While this might save on Opex it reduces the competitive operator’s scope for service differentiation, which is becoming an important factor as consumer demand for higher levels of QoS increases, driven by OTT video and also use of tablets and smartphones for browsing via WiFi in the home. The SDAN can even potentially help meet consumer expectation by distributing the control plane interface to the consumer as an app running in a PC, smartphone or tablet. This allows broadband services to be finer tuned, giving consumers the option of requesting a temporary broadband boost to watch live HD video for example.
Assia does admit that there is a need for standard interfaces to be developed for SDANs along the lines of those emerging for enterprise SDNs. There is a particular need for a standard way of exchanging messages between wholesale and retail providers to ensure end to end QoS and effective traffic management.
A number of relevant initiatives are brewing from various standards bodies, with ETSI’s NFV (Network Functions Virtualization) group examining how to move complex processing from the DSLAMs into the network, which makes sense for SDAN as part of the drive to separate control from basic transport functions. Then some of the signaling requirements cited as needed by Assia are being looked at by the ITU-T’s SG11/Q4 Q.SBAN, which addresses software defined broadband networks in general, including access and backhaul.
For SDAN to work there will also have to be a common approach to network management, so that retail operators can access relevant diagnostics and configure their virtual networks without having to defer to the wholesaler. This, as Assia acknowledges, is a work in progress, requiring agreement over security, authentication of lines and authorization for accessing appropriate data, as well as for retailers to perform control or configuration actions that may potentially affect others sharing the virtualized infrastructure.
One of the more interesting suggestions in the Assia white paper is that broadband operators could extend SDAN into the home over WiFi, or small cell cellular networks if deployed, to enable end to end control over QoS. WiFi is now intimately entwined with the broadband network and has become effectively the final access layer which therefore gives the operator potential for competing by adding value there. A cloud-based controller could assign resources such as frequency bands and time slots to femtocells, small cells, and base stations via SDAN. There is even the potential to coordinate operation across heterogeneous networks combining WiFi with mobile network that may include small cells.
We are less convinced by Assia’s suggestion that consumers will want to become more engaged with network configuration via apps. They will want the benefits SDAN can deliver, but for them to be enacted automatically as required, without having to get involved in configuration. That apart we agree with the idea that SDAN can fix unbundling for the super broadband age and ensure that retail operators can develop compelling offerings.