Capacity Crunch – stitching networks together!
Having seen the evolution of wireless from voice-centric to data-centric I can truly say that the spotlight now lies on data. And who would have thought a few years ago that the tipping point would come from apple. Apple products have done to wireless ecosystem what the warm temperatures of waters in Gulf of Mexico do to hurricanes in the gulf coast. I am an engineer by profession but all my theories of radio propagation, design principles, erlang B principles all stop to bow before the devices that are unleashed today on our networks! Wireless has become the utility like PG&E. And the pipes in networks are clogging. They are filling up faster than can be laid. Have we all become bandwidth hogs? What are we doing today that we never did in the past. One example I have from my own life – I update my Facebook status from phone, tweet every now and then, send an MMS to my circle. It is the ‘my’ profile that has changed, I used to maybe browse on my phone, send emails from Blackberry once in a while. So has the world changed faster than I did! Have the networks changed, very much too we are now on 4G – LTE/ LTE-A, is getting launched. Once SMS was the cash-cow for wireless companies, it is data now. Future networks will be networks of networks, consisting of multiple-access technologies, multiple bands, widely-varying coverage areas, all self-organized and self-optimized.
MNOs have seen a seen a multi-fold increase in data traffic. With more than 5.75 billion mobile devices in service across the globe—including 5.2 billion Global System for Mobile Communications-Long Term Evolution (GSM-LTE) devices—mobile telephony is the most dominant form of communications on the planet. Mobile devices are stoking a dramatic and unprecedented transformation in personal communications and Internet access. And wireless technology is expanding the concept of mobility and connectivity beyond the traditional phone. The opportunity for operators and their vendor partners is not just in increasing voice and data subscribers, but also in connecting every facet of a person’s technology world. Someday, the industry will look back at terms such as voice and data ARPU as legacy analytics as they consider new research terms more relevant for the all-consuming, growing connected world.
Emerging wireless networking technologies—and the ability to embed connectivity to these networks in virtually all types of devices—are creating a new connected future. While the growth of Internet connectivity globally has been well documented, it was actually constrained by fixed broadband connectivity and a PC-centric worldview. That changed with the introduction of smartphones and 3G connectivity. But unleashing the true power of the Internet actually involved moving beyond even smartphones. The tremendous expansion and evolution of wireless networks combined with growing end-user demand for ubiquitous voice and data access has helped set the foundation for a new generation of smart mobile devices, and we’re just beginning to see the potential. Yankee Group predicts that a new segment of connected devices, including enterprise machine-to-machine (M2M) connections, tablets and e-readers, will grow to more than 800 million units by 2015 (see Exhibit 1). But there is even greater potential beyond just these existing connected device segments. Some industry players, such as Ericsson and Intel, predict the connected device segment will reach 20-50 billion connections by 2020. If this seems optimistic, consider that Facebook had only 12 million users five years ago and today it boasts 750 million. Pervasive connectivity, just like the social media phenomenon, will happen. It has already begun by combining user demand, network evolution and an ecosystem that is breaking down barriers.
Broadband deployment is a central preoccupation of most if not all countries. Mobile broadband will play a significant role, particularly in developing countries, in realizing the vision of a broadband connected society. Presently, a number of countries are contemplating what strictures if any might be appropriate to impose on broadband Internet access providers related to consumer access to content, applications and services via the Internet. In doing so, many have recognized that network management allowances need to be preserved, but what that means has been less than fully articulated.
Traffic management for ensuring high quality services to consumers and overall network reliability. The importance of a basic grounding in broadband network management practices is underscored by the fact that mobile broadband networks confront a number of specific challenges, in particular that the physical layer is subject to a unique confluence of unpredictable and unrelated (i.e., orthogonal) influences.
3GPP has endeavored over the last several years to standardize increasingly more robust traffic management (QoS) techniques for mobile broadband networks such as UMTS‐HSPA and LTE.
3GPP standards acknowledge insofar as possible an E2E view of QoS, that is, from the end-user’s point of view. At the same time, traffic management needs to be interpreted in light of the fact that mobile operators typically do not have full control over E2E provision of services that depend on mobile broadband Internet access.
Radio Agnosticism and Stitching networks
Today’s operators have two, three or even technologies interworking among themselves to provide us the coverage that we need. Operators that come from a 3GPP family of technologies – GSM, UMTS, HSPA+ and LTE have backward compatibility and the ability to handover between inter-RAT (Radio access technology). And so do operators that come from a 3GPP2 family – CDMA, EVDO, LTE family of technologies. WiMAX and Wi-Fi are the disruptive technologies that have caused MNOs to adapt new avenues to offload traffic. Traditional wireless is changing as newer solutions are adapted to satisfy the hunger of always connected devices.
To allow for mobility across networks that utilize a multitude of wireless technologies, the natural first step is to enable radio agnostic handover. Many handover mechanisms today are specified to wireless technologies, e.g., WiMAX forum recommends how handover can be achieved in a mobile WiMAX network where GRE tunneling is commonly employed. These mechanisms often make assumptions about the specific wireless technologies and are not directly applicable to other wireless technologies. To reconcile the differences, we reduce handover to the lowest common denominator for popular wireless technologies, i.e., re-routing rows. To advocate flow-based management to the mobile industry is like preaching to the choir. The concept of managing the network at the row or terminal granularity is well-established in the mobile industry. However, Ethernet-IP based networks tend to manage with granularity of packets. To introduce the idea of rows in these networks, we exploit a growing movement, namely OpenFlow. OpenFlow brings the concept of a row to switches, routers and Wi-Fi APs, which can then manage packets identified to be a w from a 12-tuple that spans from Ethernet addresses to TCP/UDP ports. This allows a flexible definition of rows, which in turn provide a powerful way to manage the network using the idea of rows. This allows OpenFlow to be readily integrated into commercial switches, routers and Wi-Fi APs. By extending the simple and yet powerful abstraction of rows to Ethernet-based technologies, we were able tom demonstrate how it can be routed between Wi-Fi APs and WiMAX base-stations.
While radio agnosticism is a key component of stitching together networks, there are many other real world problems to be solved. Top three causes are:
Client networking stack is restrictive. For example, it continues to be uncommon to have multiple interfaces being used simultaneously. This is because such complex network management has to be manually configured by users today. It can be contended that a more powerful and flexible client networking stack would be instrumental in changing the way networks are used today. Maybe one day we would not need to turn on the Wi-Fi interface in Android when Skype is started.
Network today provides only a single service, i.e., connectivity. Application developers expect little more than a connection to the Internet from network providers. For example, a delay-tolerant application would not be able to query the congestion state of the network today. Sophisticated developers might even perform network probing, which in turn affects the network state itself. From our discussion with MNOs, we can see that changes in the right direction i.e., there have been talks about providing QoS services as supported by 4G wireless technologies. The willingness of MNOs to introduce new network services poises a great opportunity for researchers and infrastructure developers. We can propose and research on new network services that would help better tomorrow’s mobile networks.
IP management has to change. As we stitch together different networks, we have to reconsider the management of IP addresses used in these networks. Some networks might be NAT-ed while others provide a publicly addressable IP. Users might have a single IP address or multiple given the increasing number of network interfaces in mobile devices. Under these dynamics, session continuity is an interesting and hard problem. Even for a single service provider, they might use another’s network to serve their customers, much like Sprint using Clear’s WiMAX network shared with Comcast and Time Warner. What IP address would a customer be given in Clear’s network? Some might advocate the use of mobile IP, which would then require a scalable directory service from which the home agent of the client can be located. All these problems have to be resolved as we move towards the next generation of mobile technologies.
Here is blogpost on Wi-Fi Offloading that I had done earlier in the year.
Real world disruptive wireless
A stealthy start-up named Republic Wireless has launched based on a concept that’s enough to grab anyone’s attention, at least momentarily: unlimited voice, data, and texting for $19 a month. The company says it’s going to make that possible by routing as much stuff as possible over Wi-Fi networks, and utilizing Sprint’s cellular network where necessary.
There are several catches. For one thing, Republic will only support one phone at first: LG’s Android-based Optimus, running Republic’s custom software. (The first-month fee of $199 gets you the Optimus.) For another, the service won’t offer international calling for now. Republic cheerfully concedes these points.
But there’s another gotcha which the company’s site tap-dances around: It claims it’s offering unlimited service, but also says that it’s possible to use the service in a manner that isn’t “reasonable” and which violates a “fair use threshold.”
In one place on its site, Republic seems to say that the $19 gets you unlimited everything over both Wi-Fi and Sprint.
Can it disrupt the wireless marketplace? Well only time will tell as I am sure this business model will not work out if a customer doesn’t stay on Wi-Fi enabled networks for more than 70% of the time.