Airport PSIM / VMS RFP Examined

Airports security and surveillance systems are some of the most complex and expensive projects anywhere.

In this report, we examine a recently released RFP for the San Francisco Airport. The Airport seeks to deploy a PSIM system to integrate their various subsystems as well as a new VMS and megapixel cameras to expand the Airport's coverage. 

This is a fascinating example showcasing the challenges in building and integrating a system with this level of complexity. With numerous proprietary systems and some bankrupt suppliers, making this all work together as 'one' is an extremely difficult task. We find it ironic that privacy advocates worry about big brother when the more practical problem is simply making all this work together.

That noted, we think the SFO RFP is a relatively high quality document. The issue is less in the design than the fundamental issues that systems of this complexity face. 

Inside we dig into the details of the RFP, providing analysis and guidance on key issues and design choices.

More RFPs Reviewed

The target dollar value of this project is $3.5 Million with 9 months to complete the task.

Given the high cost (and complexity) of the project, the criteria to judge success is different than small (say under $100,000) deals. At this level, saving $100 on an individual device is relatively insignificant. The most important issues will be (1) can the bidder really integrate all these systems and make it work well and (2) how effective is the new system at improving surveillance.

Key Details

Let's review the key details before we analyze the project:

• This project requires integration of multiple CCTV systems, multiple access control systems, multiple video analytic systems, a CAD system, a perimeter system, an IDS, etc.
• The Airport will not provide any help integrating systems or requesting APIs or SDKs
• The airport wants to add a GIS enabled or a 3D command and control system / PSIM
• The PSIM/Command and Control System (called an Event Manager by the airport) shall be either Proximex, Rontal [link no longer available] or Visual Defense; The system shall include a "100% functional and useable 3-D MAP of the entire SFO campus"
• The VMS shall be either Genetec, DvTel or Visual Defense (specialist security solution targeting high end applications such as airports - 2007 revenue of $28M CAD and 104 employees [link no longer available])
• For the network, "Cisco hardware products are the only products allowed because the Airport's existing networks are Cisco based."
• Sub 300ms latency required: "Operation of the integrated systems and subsystems shall be seamless and contain less than 200 millisecond lag from activation to movement on existing and new systems"
• For determining video quality/resolution, "Design will provide proposed pixel-on-target calculations to provide general observation and tracking of passengers..."
• Integration proof must be provided: Proof of operational concept and statement of integration responsibility for all systems to be integrated including Lenel, MDI, Quantum Secure, Intergraph, Vidient, Pelco 9700 Matrix and Pelco Endura,
• Existing Pelco Endura system to be kept. No encoders, decoders or hardware can be added to integrate.
• Pelco cameras are preferred: "Pelco Analog and megapixel CCTV cameras are the SFO camera of choice. Cameras other than PELCO shall be submitted for approval by the DBI for approval prior to design."
• New cameras seem to standardize on megapixel and will use pixels per foot as the defining metric: "Megapixel CCTV cameras will be selected from 1.0 (min) to 5 megapixel (max) and selection shall be determined by Pixel on Target designs as developed by the DBI."
• "H.264 will be the megapixel recoding standard"
• Equivalent submission are allowed and will be reviewed - Bidders must submit 3 pages on the user, operational and economic reasons why the product is equivalent.
• "The existing CCTV system currently has between 50 and 60 users that are currently using a PELCO analog matrix switch. The new VMS will need to allow the existing analog users and the selected VMS workstations to override the matrix switch user and vise-versa to allow and prioritize control of each (all) the cameras." 
• The PSIM will have to integrate with subsystems on a very low level including numerous advanced functionalities and configurations.
• Extensive training: Training of 14 people (10 airport security employees and 4 TSA employees) for a full week. Additionally training of 6 facilities department employees for 2 full weeks.
• 5 Year warranty for all software including 5 years of free upgrades

Analysis

Let's go point by point:

• Key Need - Ensure Integration: With so many different systems that need to be integrated (from Pelco to MDI to Intergraph to Vidient, etc., it's unlikely that anyone can completely meet the airport's needs with their COTS offering. Securing commitments (and pricing) from both the PSIM vendor and the security system vendor is essential to ensure integration on systems not supported 'out of the box'.
• Bankrupt Providers: The RFP states that access control provider MDI is bankrupt. While we cannot confirm this independently, we called MDI's list phone number and it was disconnected. Additionally, while the RFP does not state this, we know that one of their other suppliers, Vidient is out of business. Needless to say, this can make integration much more difficult, especially if it requires any new development on the part of the bankrupt company.
• Burden on the Bidder to Integrate: The airport has made it clear that the burden on integrating these system is firmly on the bidder and that no help will be provided. We think this is prudent precaution for the airport as a bidder could try to 'escape' by arguing that some integrations are just not possible (especially because of bankrupt subsystem suppliers). On the other hand, the more the airport demands all the systems to be integrated, the more bidders will have to increase/pad their bids to deal with the uncertainly involved.
• 3D Maps: The whole airport 3D map requirement is interesting. We question how valuable it ultimately is especially relative to the cost of producing such a map. While the 2D element can be relatively easily obtained via publicly available resources, 3D maps generally require a dedicated, manual process of capturing imagery and building of the map. This could cost hundreds of thousands alone.
• PSIM Shortlist: Two of the most aggressive marketers are notably absent - Vidsys and CNL. On the other hand, the sheer requirements of this project is likely beyond their technology capabilities (though not their marketing claims). Of the companies short listed, 2 are large companies (Rontal and Visual Defense, relative to PSIM startups) and the third is a small company though based less than 50 miles away (Proximex). We think Proximex is a risky option. Given that its so small and the PSIM is such a critical component of their security, what happens if Proximex goes bankrupt (like their access control or analytics provider) or gets acquired and goes nowhere (like Vistascape)? (Perhaps Proximex will do well as recent speculation surrounding their Cisco partnership signals).
• PSIM vs Intergraph: One of the most bizarre elements of the project is that they want to integrate Intergraph with their new PSIM system. It's potentially extremely redundant as Integraph is designed to operate as command and control system for such applications. As we examined in October 2010, Intergraph is expensive ($500,000 minimum) but it won't be cheap to deploy a whole new PSIM either.
• Low Level PSIM Integration: The RFP requires a number of lower level integrations between the PSIM and the various subsystems. We think this is going to be difficult and/or costly to deliver. Most security systems only expose basic/core functionalities via API. For instance, VMS systems may expose live and recorded video but rarely do they make system configuration available.
• VSM Shortlist: The VMS shortlist makes a lot of sense to us with Genetec and DvTel offering similar overall solutions and Visual Defense providing the advantage of tight integration with its own PSIM offering. For a project with this many requirements, only a handful of VMSes are even realistic options.
• Matrix Switch Integration: In high end projects, matrix switch and keyboard integration are common 'must haves.' This RFP is no exception. The problem for VMS selection is that only a fraction of VMS systems integrate with these devices, eliminating most products.
• Moving Away from Endura: It's interesting that the airport is moving away from Endura, a product that is certainly still in its 'prime of life'. We do not know why though the airport is requiring the use of off the shelf computers/servers. This would be one element that would disqualify Endura.
• Forcing Integration with Endura: Not surprisingly, the airport wants to keep using Endura for cameras already in use and add new cameras to the new VMS. From both a financial and political standpoint, switching out Endura would be very problematic as the equipment almost certainly has many years of usable life remaining.
• Cisco Networking Hardware Required: While IP networks are based on open standards, the airport's requirement for all Cisco networking equipment is not surprising. Using a single vendor can sometimes reduce interoperability issues (especially if the network uses more advanced features). Cisco was not able to leverage its networking power into the security specification but the all Cisco network requirement is another good reminder of Cisco's wide reach in general IT.
• Sub 300ms Requirement: The airport is requiring end to end latency of under 300ms for the system. This is likely most important for PTZ controls (or two way audio if used). However, keeping latency under 500ms is generally done to make a system truly 'real-time.' On the other hand, we are skeptical how they are going to measure this (we do not see how the measurement process is defined). Depending on how you measure, you can make the latency appear to be much shorter or longer. Secondly, we are not sure what the value is of enforcing such a tight latency requirement. For instance, a fixed camera with a 1000ms (or 1s) delay is not a security risk (and may not be noticeable) even in responding to a real time incident.
• Standardizing on Megapixel: We think standardizing new cameras on megapixel makes good sense and is a important 'sign of the times.' Airports cover large, usually open areas (SFO is over 2 million square feet). In such areas, megapixel is clearly superior in providing cost-effective broad coverage. In an airport, where construction is both costly and complex to coordinate/execute, using a small number of more 'powerful' cameras is highly desirable.
• Pixels on Target Specification: While pixels on targets is a useful metric, it appears that the RFP will allow bidders to make the decision about how many pixels on target are provided. We think greater clarity on this element is needed both in terms of total pixel on target and where the target will be. Otherwise, visual detail will likely not match operator's needs/expectations. Secondly, when specifying pixels on target, the RFP should factor in any direct sunlight facing cameras (that will increase the need for WDR or more pixels) or any areas that might become dark during certain times of the day (though less likely in an airport).
• Pelco Cameras of 'Choice': We do not understand why Pelco is the cameras of 'choice.' In the old system, with Endura as the recording platform, it made some sense because of the tighter integration provided by the all Pelco system. However, integrated with Genetec, DvTel, etc. Pelco does not provide clear advantages over numerous other large IP camera manufacturers. Additionally, in the required H.264 mode, Pelco Sarix frame rate is generally low plus the cameras max out at 3MP (while the specification calls for up to 5MP cameras).
• Specifying Camera Details: In general, the camera specification is minimal (beyond the Pelco and the 1 to 5Mp requirements). Do they require interchangeable lenses, mechanical cut filters, WDR capabilities, audio, alarm inputs, multi-streaming, SNMP, etc, etc.? These elements should be clearly spelled out in a bid this large to ensure that important functionalities are not missed because their need was not communicated up front.
• Equivalent / Substitution Process: The Airport has a good substitution process where the bidder provides essentially a white paper on why an alternative product is superior. Rather than artificially force the bidder to meet arbitrary requirements, the RFP does a good job of laying out key questions to check if the alternative is capable of meeting the fundamental needs.
• 5 Year Warranty/Software Upgrades: We think their software warranty and upgrade requirement is a prudent approach. While it increases up front cost, this eliminates the risk and complexity of allocating new funds in the future. Additionally, it may reduce the cost as manufacturers often offer discounts for up front long term contracts.