The debate over mobility is in full swing. It is clear that the next frontier for the Internet is wireless. Less widely understood, however, is how mobility will impact the Internet and its applications. Broadband changed the Internet; mobility will have an even greater effect.
The transition from enterprise and dialup networking to broadband increased the data rates and load demands on Internet networks and the computing or server platforms. The increased throughput also offered software architects new ways to shift from client-server computing to truly distributed applications based on IP and Web technologies.
The coming deployments of 4G wireless networks will be the first of many generations of native packet-based wireless technologies. (Others in the pipeline include WRAN, wireless sensor networks, and various ad-hoc networks.) Past generations of cellular wireless technologies have treated Internet protocol (IP) much the same way as the public switched telephone network (PSTN) treated IP in the early days of the Internet.
Today, wireless operators with time-based call revenues see mobile data as a cash cow in much the same way that the telcos saw all of those second telephone lines being installed. Now as then, neither the telcos nor their modern wireless counterparts fully grasp the way in which software architects can rapidly develop new architectures to work around the limits of such narrowly and vertically defined revenue models (expense from the consumer’s perspective). For example, charging per byte forces developers to move applications to Wi-Fi. Similarly, we can expect video service charges to force consumers to mobile ATSC (M-ATSC).
One lesson we should all have learned is that the capability to rapidly deploy and dynamically reconfigure software will always win a head-to-head battle with any hardware and network constructed models of telecom service revenue generation.
In the space of wireless mobility, this should cause us to carefully examine the products, services, and technology choices—to build adaptable (programmable) platforms that allow service providers to adjust to new software architectures. As we continue to deploy Wi-Fi networks and begin the process for cellular network design, there are choices we can make in IP and wireless systems architecture that can position cable operators to take advantage of existing assets and to prepare for future challenges.
FIGURE 1: ATSC M/H; Source: Wikipedia (Under Creative Commons License Deed)
A brief review of the history of wireless mobility is worthwhile. The first wireless networks (radio transmission) addressed the issue of what we would now call wireless transport. That is the ability to transmit a signal from one place to another, without wires. At the same time, radio addressed the concept of broadcasting, the transmission of a signal from one place to many places concurrently. The efficiency of radio and television broadcasting is evident today as these remain highly cost-effective mass media. Satellite, cable, and passive optical networking (PON) technologies continue to use broadcast technologies because of the same technical and fiscal merits. In summary, radio provides “mobility” by broadcasting the same signal everywhere at one time.
Telephone calls are, traditionally, circuit-switched. They involve setting up a communication circuit between two parties and their respective end point equipment. Nearly all of the means of personal communication are based on broadcasting, but they have one twist. Modern cellular telephones use broadcast architectures to emulate circuit services by using codes or security measures to ensure that only the desired parties can use the selected portion of the broadcast intended for them. In the reverse (return) channel, radio signals from most cell phones are sufficiently weak that typically only the systems designed to receive the signals can process them. Here, the appearance of a switched connection is established by broadcasting the same signal everywhere (within a target cell) at one time, but only one device can decode the signal.
Not only is broadcast still the foundation for all wireless, but some types of services still make strong sense to build on broadcast technologies without the circuit emulation functions. One excellent example is the ATSC-M/H standard. This technology will allow broadcasters with licensed DTV (ATSC) spectrum to broadcast MPEG for mobile video services. In terms of spectral efficiency, the number of subscribers receiving the service is unbounded. So for any use case in which there will be more than one subscriber receiving a video stream, ATSC-M/H will be more efficient than any cellular data switched overlay. Similar cellular services can also be built using IP multicast overlays that take advantage of the underlying broadcast nature of cellular. These include DVB-H and proprietary methods such as that used by Qualcomm’s MediaFLO service.
Yet cellular is fundamentally different than the other broadcast technologies because it requires the network to track the subscriber’s location in order to support the circuit emulation model. As it turns out, that tracking requirement isn’t a liability, but potentially an asset.
Where am I? I’m here!
There are basic values upon which assets can be developed from location. These are answers to the question, “Where am I?” and the finding or knowledge of “where I am.”
The first is information that can help the subscriber or service provider provide value to the subscriber. Obvious applications are geo-location for global positioning system (GPS)-like services, E911 location determination, or the delivery of content relevant to the subscriber’s location—for example, “Tell me where the closest gas station is.”
The second is the value that can be derived from knowing where the subscriber is. The most obvious application is to direct traffic only to the cell a subscriber is currently connected to. Until recently, this has been the focus of mobility technology in cellular systems. Increasingly though, “where I am” can be of value for the service provider, the customer and numerous third parties. Today, wireless providers offer customers the opportunity to share that information with family, friends, co-workers, etc. The capability of this technology is just beginning to develop. It is hard to even imagine the limits of what can be accomplished with this technology. Some amazing examples include traffic monitoring, tracking trucks and drivers, and demonstrations of projects with Google maps.
So what do RF transmission technologies and location have to do with each other? Traditional cellular technologies need to know the subscriber’s location only for the purpose of sending data to the correct cell. Over time, that evolved to also anticipating and managing hand-offs between cells. Of course, the derivate value of location data for capacity planning and marketing developed as well.
Is it absolutely necessary to have cellular operators handle mobility? With today’s model of broadband services and Internet technology, the answer is yes. But recent software developments are beginning to show us that software can hide mobility very well. Today the only difference between software-based mobility schemes and network-based mobility schemes is that the network-based schemes are typically faster. This is in large part because of the low latency of intra-network signaling compared to software-based systems that may require client-to-server or client-to-client signaling. By definition, that latency will always be higher than signaling latency intra-network.
The distinct advantage of cable operators is not the control of or operation of mobility as a necessity, but the performance capabilities of a local network. Therein lies the challenge for cable operators. Today’s cellular wireless technologies are not distributed. They are instead centralized. They do not take advantage of the “deep” network facilities (edge hub sites) and the computing power of our edge router platforms to perform very fast and local mobility event handling to achieve high performance levels for mobility. Round trip time (RTT) reductions in the order of single to dozens of milliseconds for each message in the mobility event sequence (for mobile IP, for example) can yield significant performance advantages for an owned network vs. a cellular model with a backhaul overlay model. The edge routing platform is also the IP multicast system. It also provides security, monitoring, traffic engineering, accounting, and other functions. The cable industry’s challenge will be to develop that capability to outperform the centralized wireless infrastructures of today.
Software mobility is the operation of a continuous network service, without regard to location within a network. Whether the system provides a continuous (always on) connection is irrelevant. It’s whether the service appears to be working that way. Software technologies that replace point-to-point connections offer this kind of capability. Some technologies in use today include application level mobility (for instance, SIP), buffering, peer-to-peer (P2P) overlays, and broadcast technologies to provide the appearance of “mobility” where there is no layer 3 (IP) mobility.
Software replaces hardware
If we think about IP that would be multicast, what if software supported broadcast? Emerging P2P overlay technologies can support multicast. We can imagine future architectures that could use DVB (DVB-H, DVB-M) to stream application content. What if we even further imagined building overlay network services (like Microsoft .NET) that enabled application developers to treat multicast infrastructure like a mobile circuit switched connection?
In such an environment, the world of wireless cellular technology would face competition from wireless broadcast (only) technology. Broadcast technology with software-based coding overlays, by definition, costs less than the broadcast technology with the complex radio coding overlays. We can expect to see DVB and M-ATSC enhancements to provide exactly that functionality. Such projects are already underway with DVB-M, DVB-X, and DVB-IPTV (formerly DVB-IPI). If we welcome and endorse these technologies we can get higher data rate equivalent service to our subscribers at a lower cost.
In short, software will do to cellular networks what it has done to the PSTN. Just as with voice over IP (VoIP) (PacketCable voice), software will allow the replacement of fixed single-purpose hardware with flexible, programmable, and feature-rich software systems. In short, mobility will be programmable.
Victor Blake is an independent consultant. Reach him at email@example.com.
Sidebar: ATSC Mobile (ATSC-M/H)
Advanced Television Systems Committee (ATSC) completed the Mobile DTV specifications in late 2008. ATSC Mobile DTV (referred to as ATSC-M/H) is a candidate standard. The M stands for mobile. The H stands for handheld. ATSC-M is the North American market-specific standard similar in functionality to DVB-H. It is compatible with ATSC. Licensed ATSC (DTV) broadcasters are permitted to broadcast ATSC-M/H without further registration or licensing. The specification consists of eight parts covering overall system architecture, transmission, EPG equivalent functions, audio, video, and application environment descriptions. Each ATSC-M channel can broadcast a 416×240 line stream with video (H.264/MPEG4-AVC) and audio (HE AAC2 Vv2). (See Figure 1.)
DTV licenses for 6 MHz channels can carry about 19.39 Mbps in that spectrum. Each broadcaster can use that spectrum or data capacity flexibly. The spectrum and throughput can be divided up to allow for a range of one to eight M/H groups. The groups can be operated as a multiplex IP data stream or individually for audio/video broadcast. Two example uses of the spectrum identified by the Open Mobile Video Coalition (OMVC) are:
Network affiliate broadcaster: two to four mobile video channels and data service group(s); one to two HD channels; and two to three SD channels
Non-affiliate broadcaster: one SD primary channel; two to three SD multicasts; and 10-13 mobile video channels
Numerous other configurations are possible. For more information, visit the Open Mobile Video Coalition at www.openmobilevideo.com.
Sidebar: ATSC vs. Cellular
Although both ATSC and cellular transmission are RF broadcast transmissions, cellular RF signals are weaker because they are designed to transmit only to devices within a small radius of the antenna. This area (or cell) is dictated by the smaller of the limits of the cellular standard or design choices to shrink cells usually based on return path (uplink) performance. ATSC, like the DVB standards and older broadcast video, is designed for large radius transmission.
Although it is obvious that ATSC does not replace all of the functionality of a cellular system, it is clearly a simpler and more efficient system for video broadcast in populated areas. (The higher the population, the more efficient it is). As a rough comparison, consider the throughput capability of a 6 MHz block of ATSC (using 8-VSB) of 19.39 Mbps. If the content is broadcast to one subscriber, the equivalent data rate is 19.39 Mbps. Now consider that if the content is broadcast to 10,000 subscribers the equivalent data rate is 193,900 Mbps or 193.9 Gbps. With a short-list video lineup in a Tier 1 city, it isn’t unreasonable to conclude that 10,000 subscribers might watch a TV broadcast at a given time.
Digital Video Broadcasting (DVB) is an international consortium that develops standards as the name suggests. Many of the DVB standards are adopted and standardized by other organizations such as ETSI. The Advanced Television Subsystems Committee (ATSC) uses DVB standards including 8-VSB (which is used in ATSC, including ATSCM-H). Other common DVB standards include DVB-S used by Echostar (DishTV), elements of MPEG developed and copyrighted by DVB, and others. DVB-H and the newer DVB-SH (for S-band handhelds) are specifically for video broadcasting. DVB IP datacast (DVB-IPDC) is an overlay to carry IP on top of DVB-H. DVB-H together with DVB-IPDC is a competitive technology to ATSC-M/H. It’s widely accepted throughout most of the world. In contrast, ATSC was specifically developed for the North American market. Learn more at www.dvb-h.org/
Sidebar: Wireless RAN
Licensed DTV spectrum can be used for a variety of broadcast products and services including SD, HD, M/H (mobile/handheld) video, and to broadcast IP data. A potentially competing technology, IEEE 802.22 wireless radio access network (WRAN), currently proposed to be supported by the FCC, would allow operators to operate a service similar in technology to WiMax (based on OFDM) in the so-called white space between channels and even in licensed spectrum where it is not actively in use. Although it requires some new technology to avoid interference with licensed services, it isn’t just competitive; it could be complimentary. Since the DTV services are strictly broadcast, WRAN could offer a simple return path for IP data services. The Open Mobile Video Coalition suggests, to broadcasters, the possible use of ATSC-M/H for the forward IP transmission, and Wi-Fi, cellular, or other services for the return. Ironically, WRAN may be a better alternative in the long term.