Wireless going software defined


A problem for broadband operators and also OTT video providers is that consumer expectations are evolving faster than the service’s overall ability to meet them, which is something UK broadcaster and DTH operator BskyB conceded well over a year ago. This was in the context of pay TV in general but has quickly become equally relevant for OTT, where the bottleneck is increasingly radiating out from the fixed broadband access back into the core network, but above all into WiFi for the final hop. Even primary TVs are starting to be served by WiFi if they are web connected or using HDMI dongles and so that has become a focal point for efforts to deliver guaranteed QoS for premium services.

When WiFi is serving just one or two devices in the same room as the Access Point (AP) then performance is not an issue and is hardly any different from a direct Ethernet connection. But when the devices are scattered around a large house on different floors then usually some wired assistance is needed, using say MoCA or Powerline as a kind of backbone. There are now some robust technologies, for example from AirTies, for boosting both performance and coverage of WiFi on top of the standard, usually involving some form of mesh in which two or more APs participate as peers. AirTies has been running tests in buildings with a variety of walls and claims these show that when these are just panels dividing rooms rather than heavy structural features comprising concrete or lots of metal, it can achieve 100% coverage over up to 100 square metres with just two of its devices. This is about the size of a fairly large apartment or small house, but the mesh technology can scale up to larger areas with additional devices. When walls are thick either MoCA or Powerline can be used as a bridge to jump across. AirTies says it can share this information with operators to help them advise customers over WiFi deployment to provide the required QoS.

But operators still lack the end to end visibility extending over the WiFi network, with performance still subject to contention for example if many users are accessing high bit rate streams at the same time. As WiFi becomes increasingly ubiquitous, opera-tors would ideally like to reach into the WiFi and configure a link to a given device for a session, as some are starting to be able to do over the fixed broadband infrastructure. Here is where the emerging technology of Software Defined Wireless Networking (SDWN) comes in. This is SDN (Software Defined Networking) extended to the wireless domain, with potentially even greater scope for manipulating bit rates and QoS, but also additional technical complexity. SDN was motivated originally at the enterprise level by the drive for cost savings through use of commodity hardware and agility through being able to configure capacity for new applications faster. In the broadband access domain a motivating factor is the ability to virtualize LLU (Local Loop Unbundling) so that alternative operators can compete over a single physical infrastructure, as we explained last week.

SDN also enables differentiated QoS by separating higher level traffic control from underlying packet forwarding. In the wireless domain the ability to configure the network to optimize QoS for individual streams will be particularly valuable, because this has been largely lacking to date. In truth SDN alone does nothing to achieve this because it requires developments at the physical network level, which means here around mesh networking and distributed beam forming. The point about wireless is that performance is subject not just to pre-configured capacity but also varying factors such contention, the location of client devices and in the home changes in furniture layout or structure. Distributed beam forming attempts to overlay some order by enabling several clients to come together to form virtual antennae arrays that can then optimize their signals to target individual devices. The role of SDWN then would be to give say a broadband operator visibility and control over the configuration of these virtual arrays and in effect establish a guaranteed end to end connection with a target device.

SDWN then does have an additional ingredient over fixed line SDN, which is the active participation of the client in the process. This gives additional scope for improving coverage and QoS not confined just to cooperating as a peer in beam forming, but also participating actively in spectrum and channel allocation.

This idea emerged from the idea of cognitive radio conceived initially in the late 1990s more for the long range broadcast domain but which has subsequently evolved to embrace mesh and the newer ideas of SDWN. The idea was to allow a receiver to automatically scan channels and select the one with the best signal at a given time. The initial motivation was to support more concur-rent radio services over a given spectral band in one area. But it has since evolved into a more interactive approach in which multiple radios communicate to optimize signalling for the benefit of all and participate in mesh approaches. The end goal of cognitive radio is software defined radio where every wireless transceiver is fully configurable by a central controller and can automatically adapt its communication parameters to network and user requirements.

With the proliferation in WiFi and also data over cellular there has been a rapid increase in R&D activity to apply these ideas at the small cell level. At the end of this month a new conference on the subject is being held in Rome, the International Conference on Software-Defined and Virtualized Future Wireless Networks.

This conference will focus on the emergence of the OpenFlow protocol as a way of enabling access control from third parties in the first instance as far as the WiFi AP. OpenFlow has become the established protocol connecting the control and network routing plane under SDN with a recently established conformance testing program for vendors. So far two vendors have passed the test, NEC on the wired side and Meru Networks for its WiFi controller. No doubt other vendors will emerge with conformant offerings at the conference.

Such products represent a step towards guaranteed end to end wireless delivery by going as far as the AP, but there is then the potential to reach out further across the wireless domain by manipulating distributed beam forming, which goes beyond the latest 802.11ac standard.

Software defined wireless can also manipulate the spectrum itself for traffic prioritization, with the potential to combine that with distributed beam forming to yield temporary boosts in bit rate for a given device. It could boost bandwidth by tuning into multiple frequencies simultaneously, a bit like channel bonding over fixed networks.

But the biggest benefit will come through greater ability to cope with the real world of fluctuating bandwidth in such a way that adequate QoS is ensured. The key is that the service provider will have the ability to alter the behaviour of the “wireless last mile” as quickly as the conditions change and therefore stay in charge, rather than reacting after the event as at present.

This will be possible in three ways, firstly by dynamically prioritizing traffic over the wireless hop. This could in extreme cases even be done within a single video stream, so that for example I frames could be prioritized over B frames, because the loss of the former has a much bigger impact on video QoS than the latter. Secondly, through now being aware of the service experienced by the user, the service can adapt quickly to the user’s preferences. There is also scope for manipulating offloading between WiFi and cellular to optimize performance. The key point is that the wireless network can respond to the needs, behaviour and preferences of the end user.

There is also another dimension to software defined wireless with huge potential at the device level itself, in shrinking the number of processors needed to handle multiple wireless protocols down to one, saving on cost and power consumption. At present a typical handset has several different processors to handle a variety of protocols such as 3G/4G cellular, WiFi and Bluetooth. Since SWDN separates out the control plane and would therefore work with hardware dealing just with raw electromagnetic signals, all the wireless protocols would be elevated to software and run on a generic processor. In principle a single chip could not only handle all these mobile radio protocols, but also act as an FM radio receiver and participate in GPS tracking.

The rise of SDWN also has consequences for regulators which will most likely be discussed at the forthcoming conference, since it would require effective deregulation of the airwaves. Overall it is likely to usher in a new era of innovation in radio transmission that will also integrate wireless and wired communications more tightly.