From: Allan Angus
Subject: Comments on your recent Public Safety Communications Report
Date: September 18, 2012 2:15:33 PM CDT
To: Silicon Flatirons Center
(A Center for Law, Technology, and Entrepreneurship at the University of Colorado)
Cc: Brad Dye
As someone who has been involved in the wireless industry for several years, I thought I would offer some comments on your recent report on Public Safety Communications
( see http://www.siliconflatirons.com/documents/publications/report/201209PublicSafetyReport.pdf ).
My first experience with radio communications began in late 1969 with a community radio system called Radio Kenomadiwin. Since then, I have managed UHF/VHF land mobile engineering teams, developed digital cellular base station & mobile station equipment, designed and implemented nationwide integrated 1-way & 2-way paging systems, developed GPS and assisted GPS networks and devices, and et cetera. For several years from 1988 on, I chaired many of the subcommittees and working groups involved in the development of standards for North American cellular systems and stood as editor for many of those standards including the original ANSI EIA/TIA 553 analog cellular document. I also hold a number of patents in the area of land mobile communications and GPS. I mention this simply in order to provide some context for my comments and their potential credibility.
In the arena of public safety communications, one can generally postulate two broad modes of operation. In one, the problem has not impacted radio communications and is not likely to do so. In the other, the problem has impacted radio communications or is highly likely to do so.
Oriented as it is towards broadband systems such as LTE, your report is almost entirely within the domain of the first class of public safety issues. As such, and in my humble opinion, it almost totally neglects those significant disasters in which communications infrastructure has itself been compromised. Examples of these scenarios include 9/11, Hurricane Andrew, Hurricane Katrina, and so on. In each of these cases, the single point of failure designs of cellular networks were tested and found wanting. Any cellular system is, on the basis of its fundamental design principles, a single point of failure network since the served region is divided up into cells and those cells are served by a single base station transceiver with a single backhaul mechanism on a single antenna array and often with a single source of power. There are any number of events associated with man-made and natural disasters that can take out a base station either directly or indirectly: cutting or grounding of back haul cable or fiber, damage to cable or fiber repeaters, physical loss of antenna towers, loss of the electrical grid in the vicinity of the base station, electromagnetic pulse, etc. These events can happen due to high winds, flooding, fire, building collapse, intentional attack, and so on. Less often pointed out, but equally important in the case of an extended disaster, is the simple fact that almost no modern mobile device can use replaceable non-rechargeable batteries. If a mobile device can last only 8-12 hours on a Li-ion [ lithium-ion ] battery and the electrical grid is dead for days, all the functioning broadband wireless on the planet is of little value.
Even within the context of a functioning broadband system and a public safety event that has not compromised radio communications, your report points out many issues with such single point of failure networks. One is capacity management. There is a relatively common and yet thoroughly fallacious notion that offered load to any communications network (whether wireless or not, whether voice or data) is distributed as a Gaussian or normal distribution. In fact, almost all real offered load is distributed as some form of power law curve; that is, one with no measure of central tendency at all. Instead, almost all sources of load offer little or nothing, moving uniformly and smoothly to very few sources that offer huge amounts. Another feature of real load is that it tends to be highly correlated, and this is especially true in disaster or emergency scenarios. When an accident occurs, everyone on the freeway calls 9-1-1. When the bank robbers are making their getaway, everyone calls 9-1-1. When the CU Buffs finally score a touchdown, everyone calls 9-1-1; and so on.
My point here is that many people almost never use their mobile device. But when they do, they all do so together. This is the way of the competition for available bandwidth in emergency situations. Many networks designed specifically for tactical applications have controls for bandwidth allocation. The introduction of this sort of control on public land mobile networks yields strains between emergency responders who have a natural desire to peel off a fixed amount of access for themselves versus the public who expect to use the network they've been paying for when they finally have an important reason to use it. Tactical land mobile networks have other features not generally used in Commercial Mobile Radio Systems (CMRS); e.g., push to talk to a tactical group. This is more than just push to talk. It involves the coordination of group membership and the dynamic allocation of membership by some appropriate administrator, who may themselves have been dynamically allocated that control. A simple scenario might involve fire departments in adjacent regions. In many situations, these departments operate independently. In some situations, they will each respond to some common disaster and the coordination of their tactical efforts, and hence tactical channels, will have to be achieved under some common administration. During response, new responder teams may become involved and then leave. The implementation of access to bandwidth and control on a public land mobile network for tactical access by emergency response authorities, and automatic recognition of their chain of command, raises many and more policy issues about the fair allocation of resources and costs, let alone the design and implementation of network and specialized mobile equipment itself. There is yet another potential dark side to this issue. Should the disaster in question have any hint of man-made, terrorist or criminal activity, the immediate desire of law enforcement will be to monitor or possibly block public communication in order to gain information about a foe, real or imagined, or to limit that foe's own communication channels. This “dark side of the force” issue raises many and more policy questions again concerning the extent to which emergency responders should have their hands on CMRS networks in critical response situations.
Another issue in functioning broadband wireless systems is in-building penetration. A typical cellular system is designed to achieve a certain carrier to interference ratio (CIR) at the cell boundary. It is important for capacity reasons not to exceed this design point, otherwise the interference in adjacent cells rises to a level that compromises cellular reuse. Unfortunately, a typical building may introduce an additional loss of 10 to 20dB or more on both forward and reverse paths. If that building happens to be near a cell boundary, then achieving the necessary link power to overcome the building loss on the forward channel automatically defeats the CIR design point. Over the years, many cellular providers have compensated for this problem by placing repeaters in buildings in order to improve in-building coverage. This has been a reasonable and economic decision for large malls, office buildings, subway stations, airports, and similar structures where occupancy makes the business case. However, these in-building repeater systems are generally provided with even less redundancy than the standard base station. This implies an even greater likelihood of communications failure in situations where personnel are trapped in buildings. This would be as true for members of the public caught in a building collapse, for example, as for fire fighters in a burning building.
A related issue for emergency responders is the functionality of embedded GPS devices in their mobile units. The GPS system was designed for operation in more or less clear sky conditions with optimal sensitivity for mobile receivers with complex circularly polarized antennas capable of obtaining signals from 6 to 8 satellites. A GPS device embedded in a mobile radio using a dipole antenna and operating in an office tower is significantly compromised with respect to the original design model for military operation. As a result, almost all of these devices require some form of “assistance” so that the GPS chipset, which can consume nearly 100mA on its own when running, can be enabled only for a short interval of time when a fix is required. Such assisted GPS chips require complex network side support but provide the advantages of improved sensitivity in-buildings, improved battery life due to the short duty cycle of the chip, and improved acquisition time for the GPS satellite signal. Unfortunately, in the real world, the best of these assistance mechanisms is controlled by Snaptrack and its parent company, Qualcomm, who has a strategic technical interest in limiting access to this technology only to networks where it earns significant patent royalties. It should go without saying that it can be critical to locate an “officer down” in an emergency situation in the most accurate and timely manner possible. That access to the technology to do this is limited by some organization's patent policy is troubling to say the least, IMHO.
There have been, and remain, good solutions to these problems in systems designed specifically for emergency and tactical applications. Such systems build in redundancy both within the network and at the air interface. A common mechanism to achieve network redundancy is simply the use of multiple elements of equipment and backhaul to limit the possibility of any single point of failure from taking out the network. Likewise, redundancy in the air interface can be achieved through simulcast mechanisms or the provisioning of significant overlap in coverage or both. As well, the use of multiple base station receivers to provide reverse channel redundancy is a natural mechanism. In-building penetration is achieved by avoiding cellular style systems in which carrier to interference is a design factor and working instead with dedicated bandwidth allocation schemes in which carrier to noise ratio is a limiting factor. On the mobile side, operation on replaceable non-rechargeable batteries is essential in the case of any extended disaster scenario. The ability to transmit true broadcast traffic on the forward channel to a multiplicity of devices without incurring any significant bandwidth issues is of great importance in communicating common traffic both to the public and to emergency responders.
While almost any individual who has become interested in mobile communications over the last 10 years will consider what I am about to say truly laughable, almost all of these attributes have been embodied in paging networks. It goes without saying that the common view of paging is that it is an archaic technology, justifiably headed for the junk heap. The fact of the matter is that the most recent round of 2-way paging networks, e.g., ReFLEX and InFLEXion from Motorola and pACT from ETSI, were designed in the late 1990s and incorporate many of the lessons learned from digital cellular systems such as the GSM, CDMA, and so on. As well, and oddly enough, Snaptrack assisted-GPS has been licensed to operate on such networks although the licensee is engaged in a law suit with Snaptrack and Qualcomm and has suspended operations at least until that suit is settled. [I should point out that I personally have an interest in the outcome.] It is true that these are narrowband systems that were never designed to offer streaming video; for example. However, it seems to me that these specialized designs are far closer starting points, in terms of satisfying the broad set of requirements of first responders than what could ever be delivered on CMRS broadband systems strictly on the basis of their inherent design criteria.
The bottom line on this set of comments is that it would probably be a more sound approach to begin with a set of networks whose design goals have yielded results that closely satisfy the essential elements of a public safety network, and then improve their bandwidth, than to begin with a network that has the desired bandwidth and delivers on none of the essential elements of a public safety network.
Just saying. . .
P.S. I have copied these comments to Brad Dye, a long time friend, who publishes a newsletter for the critical messaging sector.