Matching access resources to service requirements at the lowest cost per bit is the key to differentiation, says Andrew Mackay, Cisco's manager for mobile architectures for Asia-Pacific, Japan and China
TEMEA.net: What is driving telcos toward carrier-grade Wi-Fi offerings?
Andrew Mackay: First are the challenges that mobile network operators (MNOs) are facing with the proliferation of large-screen mobile devices with internet-based multi media applications. The networks are struggling to meet the rampant demand in mobile broadband connectivity and bandwidth. At the same time, mobile data revenues have failed to keep pace, leading to rapidly declining profitability. Here Wi-Fi has become attractive because it can add additional access capacity at a lower cost per bit.
Second is the changing nature of the mobile value chain. MNO's traditional services are under threat from internet based "over the top" services. Wi-Fi is also being looked to as a way to diversify the business model, from just consumer and enterprise offerings to business-to-business offerings, such as wholesales access, machine-to-machine, connected venues and third-party service enablers like location analytics.
How can they best deal with spectrum scarcity and constraints in macro site acquisition?
There are three ways to increase wireless capacity density: add more bandwidth (more Hz), increase spectral efficiency (more bps/Hz) or reduce the number of users per radio channel (more bps/Hz/user).
New spectrum we know is scare, with limited new allocations above 2.3 GHz and 700 MHz (digital dividend) in APAC countries over the next five years. This is why service providers have turned their attention to the use of unlicensed spectrum, of which in most countries there is over 400 MHz in the 2.4- and 5-GHz bands. Compare that to the average of 60 MHz MNOs hold of licensed spectrum.
Spectral efficiency has improved with every new generation of mobile technology, but 4G technologies like LTE and LTE-Advanced are close to the Shannon limit (the theoretical maximum data rate a communication channel can support). So the bulk of wireless capacity expansion is going to come from more sites, and since macro-site acquisition has become very slow and costly, an underlay of smaller cells to soak up capacity hotspots makes sense.
How can operators offer differentiated access across licensed and unlicensed spectrum to reduce the cost per bit?
Service providers know they have to be able to control the quality of experience for their subscribers. The end-user doesn't care if they are on licensed or unlicensed spectrum, they just want connectivity for their services whenever and where ever they are - a requirement we term "Data in Motion". So the secret to differentiation is utilizing whatever access method is available that offers the minimum service requirement for good QoE.
The secret for profitability is utilizing the access that meets the minimum service requirement and offers the lowest cost per bit. Hence the concept of a heterogeneous network, one that has multiple disparate access technologies, each with different service capabilities and cost-to-service characteristics. Of course this "tool box" approach is useless without the orchestration to match the required service and cost constraints for a session on the network. That is where the intelligent core comes in.
Can you define a "common converged packet core" for heterogeneous access?
A common converged packet core is the heart of a heterogeneous network. It is the central point where access networks of different technologies and topographies standards converge and hence becomes a control point. It is also the anchor point for centralized services, analytics, policy and charging.
We are only now starting to see true convergence in the core. A mix of standards development and vendor innovation means a true multi-access core is possible. Cisco for example has a packet core solution built on software virtualization. This allows the same hardware platform to support core functions for 2G, 3G and LTE radio access, plus Wi-Fi and femtocells, supporting open standards based interworking for 3GPP, 3GPP2 and IEEE interfaces.
How does intelligence embedded into the network address operators' needs and what are the key benefits?
I mentioned before that matching the available access resources to the service requirements at the lowest cost per bit is the secret to MNO differentiation and profitability, and that is the definition of intelligence. For the network there are three element of intelligence. First is analytics to understand the minimum service requirements, and available access resources. Second is the computation to assess current service capabilities and the associated "cost" available resources. Lastly is the orchestration to direct the device and network to the optimal exchange. Obviously, this is no trivial process with a complex interaction between radio resource management, traffic inspection, analytics, plus network and device policy. As you move to different cell layers and different technologies it becomes even more complex.
How do telcos go about implementing that and how does this solution differ from what other companies are offering?
Recognizing the complexities of "network intelligence", the broad aims of development in this area, is around simplification and abstraction. By simplifying the complex interactions between analytics, computation and orchestration, service creation can be made faster and more dynamic; and by abstracting the intelligence to third-party applications new business models are enabled.
Cisco has designed a network abstraction layer that achieves these aims. Called Mobile Internet Network Enabler (MINE), it unifies the intelligent network functions in real-time and exposes the control to high-level applications and services without customization. This framework is part of Cisco's vision to an open, programmable and application-aware network.