Wireless operators in the USA face a fiercely competitive market. Growth in new subscribers is slowing as the North American market is nearing complete penetration. New market share must be captured from the competition which is a much more costly endeavor than acquiring new (to the wireless market) subscribers. Competition is also putting pressure on pricing - ARPU (average revenue per use) is flat to declining.
The growth in new services, demanded by users, adds to the wireless operators’ challenge as well. Previously, voice and text messaging was a full-offering. Today’s users are demanding interactive services such as high-speed internet access, multi-media (video and music) capabilities, and rich mobile content (maps, social networks, targeted advertising) not imagined several years ago. These services share one common attribute – all require significantly more bandwidth than is currently available.
With nominal revenue growth, but significant up-side in bandwidth hungry services, network operators are focusing their attention on infrastructure (OpEx) efficiencies; how to both deliver additional services and cut operational costs. The backhaul network has now become the bottleneck to growth!
Traditional backhaul networks consist of multiple T1’s lines leased from the LEC (local exchange carrier) feeding each cell tower. With just voice services, there was a direct relationship between the amount of revenue (billable minutes) and the required number of backhaul circuits. Capacity planning and budgeting was easy in this model.
But these new mobile services are packet-focused, very bursty, and require significantly more bandwidth than can be provided by a T1 based network. Operators are evaluating several options to improve scalability, quality of service, and cost-performance.
Ethernet Offload (hybrid model)
Multi-Service SONET/Optical Networking
(Ethernet / IP / MPLS) Networking
To meet the increasing demand for packet based services, a number of operators are building/leasing packet (Ethernet/IP/MPLS) overlay networks. The separate data transport network provides native packet hand-offs at each cell site, and provides best effort transport for services such as internet access, multi-media downloads, and interactive networking services in a very cost-effective manner.
| Pros | Cons |
|---|---|
| No QoS issues with voice (still TDM based) | Double OpEx – managing two networks |
| Minimizes service disruptions | No elegant migration of traffic |
| Limits growth in expensive leased lines – T1’s | |
Leasing data lines limits CapEx |
|
The experience with this model has been mixed. Although some operators have experienced good ROI using this model, some have found the operational expense of managing two networks outweighs the incremental savings of using the cheaper data lines. Regardless of experience, the strategic view is that this architecture is a short term “migratory” approach toward a final “all packet” solution.
One of the most trusted and proven technologies in the market – SONET – can provide an excellent backhaul solution for mobile operators. “Next generation” SONET ADM’s transport TDM traffic natively, providing a platform that incurs no packet loss, limited jitter, and minimizes latency across multiple nodes in the ring.
The “next generation” designation indicates a platform that also incorporates significant support for data services. These ADM’s incorporate GFP for a standards based, low-latency transport of Ethernet over SONET. VCAT enables incremental management of total throughput for dedicated Ethernet services, and LCAS allowing provisioning engineers to modify this total throughput while in service. These capabilities, combined with spanning tree enhancements and RPR options provide a true, carrier class Ethernet transport platform capable of meeting the strict latency and jitter requirements presented by cellular data backhaul.
The newest solution for cell tower backhaul is a pure packet-based transport. This solution leverages the costs-effectiveness and transport-efficiencies of native data protocols such as Ethernet and MPLS in order to provide a carrier grade, yet very inexpensive backhaul network. Making this solution work requires significant enhancements to existing data gear.
First, the data network must support some form of TDM circuit emulation or tunneling protocol. Because much of the existing gear in the RAN today does not support native data interfaces, a combination of data and TDM circuit needs is expected to exist for several years to come. Circuit emulation capabilities in the network enable a converged, single infrastructure, and facilitate a gradual migration to a total-packet solution. Today there is a thriving market for Pseudowire gateway units purpose built to meet these needs.
Secondly, base-stations (located at the cell tower) need some form of synchronous timing, in order to cooperate effectively while handing off calls between towers. Recent advancements to Ethernet (synchronous Ethernet, PTP) are providing accurate timing sources and clock recovery capabilities to meet these needs.
Finally, the data network must support extremely rigid jitter and latency parameters in operation. Success of packet backhaul solutions, in great part, depends upon the resiliency and tolerance of the RAN gear. To date, packet solutions can meet some, but not all of the latency/jitter requirements being put forth by wireless operators. Backhaul providers (or operators looking to build a packet backhaul network) must carefully consider when and if a pure packet backhaul strategy will be feasible considering their specific network QoS parameters.
