Public Safety Networks Aren’t Special
The White House released a paper today called “The Benefits of Transitioning to a Nationwide Wireless Broadband Network for Public Safety” arguing for a single–purpose mobile network for public safety. The paper addresses the options that come from the desire to improve the communications capabilities of first responders and the set-aside of dedicated RF spectrum for the exclusive use of public safety. Wisely, the paper begins by recognizing that the typical American teenager has better communications capability than the average cop:
At a recent hearing before the Senate Commerce Committee, New York City Police Commissioner Raymond Kelly remarked that a 16-year-old with a smartphone has “more advanced communications capability than a police officer or deputy carrying a radio.
It then inexplicably argues for a continuation of the policies that created this condition. The essence of the White House’s mistake is a failure to comprehend the capabilities of the all-IP LTE network. As is common in debates over this subject, the paper makes too much of the supposedly unique requirements of public safety:
The traditional LMR systems and devices developed for public safety have served public safety agencies well with regard to meeting their unique requirements. Most notably, such systems are developed to provide rapid voice call-setup and group-calling capabilities. (Ordinary cellular systems, by contrast, can allow for seconds to go by before a call is delivered and answered.) When time is of the essence, as is often the case when public safety agencies need to communicate, it is important to have access to systems that achieve fast call-setup times. Similarly, unlike ordinary cellular systems, dispatch systems like those used by public safety allow for large talk groups to communicate either among individual units or by broadcast messages (think: “calling all cars”).
Above and beyond rapid call-setup and group-calling capabilities, public safety agencies also depend on a number of other important functionalities. Most notably, public safety relies on devices that allow for a handset feature known as “talk-around,” which enables two or more mobile or portable units to communicate without the aid of network infrastructure. In the case of emergency situations where such infrastructure is not available, a peer-to-peer mode of communications is crucial. Similarly, modern public safety dispatch networks provide queuing and priority access capabilities that traditional cellular networks were not designed to provide. In short, despite their operability and interoperability limitations, traditional LMR systems have provided public safety agencies with mission-critical capabilities that conventional cellular systems have not generally offered. These systems will continue to be essential for public safety communications until broadband systems are able to meet public safety requirements, particularly for mission-critical voice.
Any competent IP engineer should realize that these “advanced” capabilities are achievable simply as apps for any good smartphone platform operating on an IP network and don’t require special networks or special handsets. Let’s see what’s involved in priority access, quick call connection, group calling, and network bypass (“talk around.”)
- If the cops need “priority access” they can use the IntServ capabilities specified for LTE in the IMS standard, or they can use garden-variety DiffServ. Preserving the ability to run different apps at different QoS levels was a big part of the net neutrality fight, and thankfully mobile nets got an exemption from the presumptive ban on advanced networking capabilities in the wireline world.
- The quick setup (walkie-talkie) thing is a trivial example of VoIP running over UDP or even TCP. The setup time for a UDP connection is zero, since UDP assumes that setup is omnipresent and doesn’t need connections. All you need is a network that can route to an IP address and a system that binds IP addresses to phones. We’ve only had those things for 20 years. I’m seriously astonished that anyone can say that auto-answer VoIP is a “special app” that justifies a unique network.
- Group calling is a also a straightforward VoIP application that millions of people use everyday from a variety of commercial services. Like conference call bridges, each user connects to a conference hub and that hub makes a group call happen. You predefine some groups, assign them to buttons, and there you go. This capability is simply and end-point service on an IP network, not a specialized phone or a unique network. Alternately, this service can be provided by IP multicast without a central server.
- “Talk Around” is actually a means of bypassing a non-functioning network, so by definition it doesn’t argue for a special network, although it may require a special phone in some sense. It’s going to be implemented as an application in any case. There is a similar “peer to peer” mode in Wi-Fi that enables two or more devices to communicate with each other, so there’s an argument that it’s already done. Wi-Fi has limited range, so we’d want to extend that a bit by using LTE spectrum. The sensible way to to this is to simply migrate some of the code that makes P2P Wi-Fi work (it’s software, not a special chip) to run over the LTE radio. That’s not particularly easy, but the work that needs to be done is pretty much the same whether you put it in a special phone or on a high-end iPhone or Android.
There are some fairly refined policy arguments for allowing public safety to operate their own networks based on conflicting priorities during times of crisis and making the most of limited spectrum. The requirements listed in the White House paper don’t address the subtle points of policy, they’re direct technical claims about the capabilities of LTE networks that range from false to irrelevant.
It doesn’t matter how long the cellular network takes to connect a call when your technology target is LTE. LTE is an all-IP network on which calls are going to be some form of VoIP. IP supports prioritization, quick connections, multicast, and P2P communication, all the things that public safety says they need. What’s missing are the apps that unlock these capabilities, and every one of the apps is something that will ultimately be done by the private sector anyway.
If we go down this road, in 5 or 10 years, there will be another Congressional hearing in which a lead witness will say that a 16-year-old with an LTE Advanced smartphone has “more advanced communications capability than a police officer or deputy using the public safety network.”
There is no sound technical argument for a special-purpose mobile network.
Wouldn’t a dual-SIM (or GSM + CDMA) phone that can connect to two carriers offer more redundancy that a single purpose built network? It costs tens of billions of dollars build a nationwide cellular network and it will be difficult and extremely costly to tax payers for government agencies to replicate that infrastructure. Why not leverage two or more existing carriers using commodity hardware and software and not try to reinvent a $20 billion wheel.
Whenever there was a major disaster in recent years, emergency workers always fall back on their cell phones because their specialized boutique gear didn’t work with each other. Yes I know they are working on inter-operable gear for emergency workers, but even those have issues inter-operating.
Commercial wireless networks can be configured to give emergency workers all the priority they need. A pooling of spectrum and infrastructure would be beneficial to everyone.
Yes, they can be configured. I’d expect some tweaks would be needed in the air interface for the talk around mode, but I wouldn’t see it as that big a deal.
Moreover, the issue can be looked at more holistically as increased robustness in the event of degraded network conditions. So, in addition to talk-around you would like the handsets to be able to act as a base station or a relay station (relaying between a user and a base station). One might also like the base stations to be able to perform a backhaul function over-the-air on mobile broadband frequencies in the event of loss of backhaul. IEEE 802.16n is looking at these and other network robustness issues for WiMAX, which is similar to LTE. These would be more-expensive, non-consumer handsets, at least initially. It’s doable, once the network architects get over the shock of seeing two mobile stations off to the side connected to each other, but not to anything else. (They HATE that.)
Mobile to mobile communications can cause huge problems in interference. Cell networks are carefully configured and managed and the large towers offer great coverage.
It’s not just some irrational architectural bias against client-only mesh networks because they have some fundamental problems http://www.digitalsociety.org/2010/07/problem-with-mesh-networks/.
The public safety mesh is for use in disaster scenarios where the backhaul network is down and there’s no concern about interference, George.
Yes I understand that, but I was referring to the common notion out there that mesh is a replacement for command-and-control architecture and cell carriers. It was in response to this comment:
“It’s doable, once the network architects get over the shock of seeing two mobile stations off to the side connected to each other, but not to anything else. (They HATE that.)”
I think Steve said that mesh-like functions are a supplement to tower-based networks in certain conditions, which is a reasonable assessment. There are a number of fringe scenarios in which meshes and mesh-like configurations are useful, even though mesh as a whole doesn’t perform well in high-density configurations.
In the 802.16n meetings there were several concerns and objections to a direct mode based on interference, how the mobile to mobile connection gets established without a base station, maintaining time sync, short mobile to mobile range, how they rejoin the network when a base station is reestablished, etc. The mobile broadband folks aren’t used to thinking like that, but they can and did. It was mostly culture shock, and I suspect a little IPR jockeying.
The amendment text is still being drafted but current text has no autonomous direct communication if the base station is working, which addresses one interference concern. If the base station goes down, all the mobiles sense this, and the first one choosing to do so can be the coordinator, acting as a base station for frame-synchronization purposes by occasionally sending a preamble for the other mobiles in range to use. More distant mobiles not in range of the coordinator might set sync using some form of ranging signal exchange yet to be worked out. The coordinator doesn’t do anything else. The mobiles in this mode will use some form of contention-based protocol for access.
In discussions leading to the completion of the requirements document for the amendment, there were separate proposals for tree and for mesh architectures, with neither being included upon the realization by the group that requirements documents should not propose solutions. When the amendment is done, everyone will be free to look at it and call it what they want.
[…] dollars of redundant infrastructure. Richard Bennett explained why Public Safety networks really aren’t so special and why the government network could be better served with existing and new commercial wireless […]