RESTRICTED
Lantern
Project options for pilot
extension
and national capability
REFERENCE:
IDT004–118-0302-Project options national rollout
AUTHOR:
VERSION NUMBER: 0.3
DATE:
21 06 2007
© NPIA (National Policing Improvement Agency) 2007
The copyright in this work is vested in NPIA (National Policing Improvement Agency) and the information
contained herein is confidential. This work, either in whole or in part, must not be reproduced or
disclosed to others or used for purposes other than that for which it is supplied, without PITO’s prior
written permission, or if any part hereof is furnished by virtue of a contract with a third party, as
expressly authorised under that contract.
RESTRICTED
RESTRICTED
Lantern
Project options for pilot extension and national capability
Product Control Page
Authors
:
Address
National Policing
020 8358 5512
:
Improvement Agency
07795 667737
New King’s Beam House
22 Upper Ground
London
SE1 9QY
Distribution List:
<names>
<names>
Sign Off Authorised By:
Name
Responsibility Signature
Date
Issue Control:
Version Date
Description
0.1 07.05.07
First
draft
0.2
21.06.07
Reviewed and amended by MC
0.3
22.06.07
Reviewed and amended by BT
Controlling Documentation
Source documentation used to produce this template includes:
•
Version 0.3
RESTRICTED
Page 2 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
Table of Contents
1.
BACKGROUND ....................................................................................................................4
2.
EXTERNAL CONSULTATION .............................................................................................5
3.
PURPOSE OF THIS DOCUMENT.......................................................................................5
4.
METHODOLOGY ..................................................................................................................6
5.
PILOT EXTENSION BEYOND DECEMBER 2007............................................................8
6.
CURRENT PILOT USAGE .................................................................................................10
7.
NATIONAL SOLUTION PROCUREMENT OPTIONS......................................................14
7.1 NATIONAL PROCUREMENT OPTION 1 – Keep same device and provider and expand
to all forces...................................................................................................................................16
7.2 NATIONAL PROCUREMENT OPTION 2 – Keep the Fingerprint Database with IDENT1,
and use devices provided by other suppliers...............................................................................18
7.3 NATIONAL PROCUREMENT OPTION 3 – Fingerprint Matching service and devices
provided by independent companies. IDENT1 provides database with ‘raw’ fingerprint data. ...21
8.
PRINCIPAL KEY ISSUES OF ANY NATIONAL SOLUTION..........................................23
8.1 PRINCIPAL KEY ISSUE 1 - DEVICE SIZE ....................................................................23
8.2 PRINCIPAL KEY ISSUE 2 - ACCURACY ........................................................................25
8.3 PRINCIPAL KEY ISSUE 3 - IMAGES AND ENCODINGS (TEMPLATES) ..................27
8.4 PRINCIPAL KEY ISSUE 4 - COMMUNICATION METHODS........................................29
8.5 PRINCIPAL KEY ISSUE 5 – FUNDING ..........................................................................30
9.
PILOT EXTENSION - OPPORTUNITIES TO EXPLOIT UNDER USAGE.....................30
APPENDIX A - COSTING ASSUMPTIONS .............................................................................30
APPENDIX B - DECISION MAKING TREE – LANTERN OPTIONS....................................30
Version 0.3
RESTRICTED
Page 3 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
1. Background
The Lantern project refers to the capability of providing mobile fingerprint
identification to the officer on the street.
It essentially provides the same functionality as that provided by the ‘live ID’
function of the custody based ‘Livescan’ fingerprint machine.
As yet the device does not connect to any other database other than the IDENT1
fingerprint database. Advantage has been taken of the demographics stored
against the fingerprint files on IDENT1 ( File name, Date of Birth, Gender), to
enrich the return made to officers in addition to the CRO number.
The device uses GPRS technology to connect to IDENT1 directly using a Virtual
Private Network which is encrypted to protect data integrity. Results are sent
directly back to the device and its user.
At present there are 10 forces using 100 devices.
Bedfordshire, and West Midlands who were used as lead forces, along with
Lancashire, West Yorkshire, North Wales, Essex, Hertfordshire,
Northamptonshire, British Transport Police, and Metropolitan Police.
The devices have not been distributed equally between forces, and this has been
dealt with in previous papers.
The current pilot is due to run until 12th December 2007.
The pilot is contracted to Northrop Grumman as a change control note (CCN)
against the main IDENT1 contract. In that CCN an option to extend the life of
the pilot for a further 12 months (subject to additional cost) has been included.
The CCN stipulates that Northrop Grumman require at least 60 days notice to
implement this extension. This means the last date this can occur is the 13th
October 2007.
It is important to note at this stage, that on the 12th December 2007, if
no extension of the contract exists, the matchers and the connectivity to
IDENT1 will be turned off, thus the police service loses what limited
capability for mobile fingerprinting it already has, at that moment.
Version 0.3
RESTRICTED
Page 4 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
2. External Consultation
In order to solicit the views and opinions of the police service, a user conference
and workshop was held on 10th May 2007, which involved the 10 pilot forces
and other identified stakeholders. This conference represented roughly 20% of
the total number of police forces in England, Wales and Scotland. Large
metropolitan forces along with small, medium, and large county forces, all being
represented.
Both users and those from a senior, strategic perspective from each of the forces
and stakeholders, attended the conference and a number of workshops were
held on the various options discussed in this document.
The output from that conference can be taken as being widely representative of
all the forces across the country, and as a result it has been incorporated into
this document to illustrate the wishes of the people using the service.
The outputs of those workshops have been considered in the preparation of this
document. However, two clear messages came out loud and strong.
• The police service would like to see the introduction of the mobile
fingerprint identification capability to all forces as soon as possible.
• Start with a basic service and build on it. Don’t go for the ‘all singing and
all dancing’ version at the beginning.
3. Purpose of this document
The purpose of this document is to identify and examine the various options that
are available to the project board, to continue and develop the existing pilot
capability into a country wide, national, service.
In considering the options available, it has been identified that because of the
combination of choices available, the options numbers could be colossal when
taking into account all possible permutations.
The methodology used in preparing this paper is explained below.
Clear principle factors in deciding which option to approve, are going to be
timescales and funding.
The implementation of any decision made on the future national solution of a
mobile fingerprint identification system will be determined by the option chosen.
Each of the options examined in this paper have implications in respect of
funding and time it will take to deliver.
Version 0.3
RESTRICTED
Page 5 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
It is important that when considering the options available that the expectations
of the police service for a system to be delivered in the shortest time possible,
are tempered against the need for securing funding, not to mention the
feasibility of delivering the necessary components to make it viable.
In that respect it is also important that the expectations of the government
(Home Office), the NPIA, and the police service in relation to timescales to
achieve the national solution are examined and an agreement reached on what is
required.
This document will now deal with a number of issues.
The first issue will be the extension of the ‘pilot’
The second issue will be looking at the various options for the national
solution, with the expectation that a preferred option will be identified to
give clear direction to the project team.
The third issue will look at options to exploit the current under use of the
pilot central checking capability.
4. Methodology
It is necessary to ensure that those tasked with the decision making, are aware
of the links and dependencies each option has, and what impact one decision
may have on choices available further down the line. For example, if a decision
‘A’ is taken it may preclude the choices ‘X’ and ‘Y’ further down the line from
being viable.
In order to best understand the various options, and their connections to each
other, documents have been prepared to help the reader follow the logic and
the explanations given in the form of flow diagrams or decision making ‘trees’.
These additional documents should be read in conjunction with this narrative.
Essentially three broad options have been identified, and these are examined
individually.
When considering the options, there are 5 clear principle issues which will affect
the choices available. These have been identified and are examined individually.
The key issues are essentially technical in nature, and it may be that there is no
need to opt for a specific solution, letting market forces deliver what is needed.
However it is necessary to understand the implications if certain options are
chosen in the timescales and costs that will be necessary for that option to be
delivered.
Version 0.3
RESTRICTED
Page 6 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
Essentially these principle issues can be correlated to specific branches in the
‘decision making tree’ (Appendix B).
An attempt has been made with each of the options to give an indication of
expected costs, and timescales for procurement and implementation. It is very
difficult to tie these down, as the choices made for the principle issues could
have a dramatic effect on time and costs.
In all cases the minimum time and cost has been identified. Once the principles
of the way forward have been agreed detailed planning can commence.
Each of the options can be readily assessed against the impact the option might
have, in delivering against any overall time and cost constraint that may be
imposed.
The project team are keen to adopt a method of “Evolutionary Development”.
This essentially means that the project delivers a simple, workable solution that
can be taken into use by the police service, whilst in parallel develop and deliver
as ‘upgrades’ more advanced functionality and integration.
The reason for this approach comes from lessons learnt on other national
projects. Users in the police service have been critical in the past that too much
time is lost trying to deliver the ‘all singing, all dancing’ version, thereby missing
the boat, and earning the phrase “delivering yesterdays technology today”.
Costings and Timescales
Costings and timescales indicated in this paper can only be taken as being
indicative rough orders of magnitude, as a detailed analysis has not been
undertaken. The exception to this is the pilot extension cost which is set out in
the original CCN.
Current Usage
Understanding the current pattern of use of the Lantern system is also likely to
assist in the decision making of identifying the best option.
Thus the current usage of the system has been examined and later on in this
document the results of that examination have been presented for information.
Version 0.3
RESTRICTED
Page 7 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
5. PILOT EXTENSION BEYOND DECEMBER 2007.
This is the most important and pressing item to be considered by the project
board.
The current pilot is planned, and funded, to last until the 12th December
2007.
At that point, the devices will be taken back, the GPRS / CJX connectivity turned
off and the matching capability within IDENT1 re-used elsewhere.
Should this be the case, a hiatus will occur between the end of the ‘pilot’ and the
commencement of any national capability.
This could be potentially damaging to the police service, the credibility of the
NPIA, and to the outcome of any possible national solution.
It is unlikely that any course of action to deliver a nationwide capability
will be achieved by the end of 2007 and such a hiatus should be avoided
at all costs.
As previously identified, in negotiating the change control notice (CCN) for
Lantern on the IDENT1 contract, an option to extend the pilot for a further 12
months was inserted, with an indicative cost. To prevent any hiatus occurring
this option MUST be exercised.
Pilot Extension – Stay as is – 10 forces 100 devices – Points to consider.
This option would see the current ‘pilot’ continue in the same vein for a
further 12 months.
This will continue to restrict availability of the devices to the existing 10
forces.
There have been a number of requests to deploy the devices to operations
being conducted by forces who are not part of the pilot. One or two have
been supported but this has met with some contractual problems
regarding device numbers, and technical support. Resolving and
supporting these requests has taken NPIA project staff away from their
normal work. As Lantern successes continue to mount, it is unlikely that
the project team, nor pilot forces acting as mutual aid, will be able to cope
with the level of demand.
By extending the pilot for another year, this will allow time to identify,
develop and implement, a national solution and roll out.
Version 0.3
RESTRICTED
Page 8 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
COST:- (subject to confirmation)
£ 386,000 (+/- 15%)
TIME FRAME TO IMPLEMENT :-
BY 13th October 2007 the
10 pilot forces would continue as they are now. This would
all happen in the background.
RECOMMENDATION 1
Negotiations should be commenced with Northrop Grumman
immediately to at least extend the current pilot on an ‘as is’ basis
until December 2008.
This is to be done as soon as possible in order to meet the 13th
October 2007 deadline.
Version 0.3
RESTRICTED
Page 9 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
6. CURRENT PILOT USAGE
To assist in the decision making processes it is beneficial to understand
the current usage profile of the system.
Analysis of the usage during the current pilot has been carried out by
Northrop Grumman. Although primarily to establish the scalability issues
needed for any potential full roll out to all forces, it also provides valuable
information on the viability of any options that may be considered in the
short, medium and long term.
The following three graphs illustrate their findings.
The period used for analysis is the 1st February to 24th April. This
represents the period from the last force having rolled out, to a suitable
cut off point to enable Northrop Grumman to produce an interim report. A
full report will be delivered to the project by the end of July 2007.
The project team felt it would be helpful if the information contained
within the report were made available to the project board to assist in the
future options deliberations.
Daily usage 1st Feb to 24th April 2007
80
7 0
6 0
5 0
s
quest
e
4 0
a
r
ch R
e
S
3 0
2 0
1 0
0
r
r
r
r
r
e
b
b
e
e
b
eb
a
r
a
a
a
a
p
r
p
r
p
F
F
M
M
A
-
A
A
1-
8-
-
F
-
F
1-
8-
-
M
-
M
-
M
5-
15
22
15
22
29
12
19-
D a te
Version 0.3
RESTRICTED
Page 10 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
12:00 AM
11:00 PM
250
1:00 AM
10:00 PM
2:00 AM
200
Times of usage across
9:00 PM
3:00 AM
24 Hours.
150
8:00 PM
4:00 AM
100
7:00 PM
5:00 AM
50
6:00 PM
0
6:00 AM
5:00 PM
7:00 AM
4:00 PM
8:00 AM
3:00 PM
9:00 AM
2:00 PM
10 00 AM
1:00 PM
11:00 AM
12:00 PM
Monday
500
450
400
350
Sunday
Tuesday
300
Usage on
250
Days of the Week
200
150
100
50
0
Saturday
Wednesday
Friday
Thursday
Version 0.3
RESTRICTED
Page 11 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
It should be no surprise that the pattern of usage is mainly during weekdays and
during the day. This reflects the pattern of deployment by most ANPR units, who
are the principle users of the device.
The Pilot system was set up to handle 200 searches per hour and to achieve a
response time of under 5 minutes.
The response time has easily been achieved during the pilot with most responses
being completed within a 2 minute time frame. This is most probably because
the volume of checks being carried out is no where near the capacity planned
for.
When looking at check capacity, two things have to be borne in mind – firstly
that the checks are never going to be evenly spaced out across the 24 hour
period, and secondly there are likely to be ‘bursts’ with several transactions
arriving at or near the same time.
The graph below shows the frequency of search numbers per hour. The bulk of
checks appear in the 12 to 16 checks per hour bracket, with occasional peaks of
20 -22.
120
100
80
e
s
nc
c
curre
60
e
r
of O
b
Num
40
20
0
12
14
16
18
20
22
24
Number of Searches within an Hour
Version 0.3
RESTRICTED
Page 12 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
Northrop Grumman found it extremely difficult to analyse with any certainty the
‘bursts’ that occur from time to time.
What they did find was :-
A number of occasions occurred where there were bursts of 3 to 5 checks
arriving with in a short space of one another. Unfortunately there appeared to be
little consistency in the outcome.
For instance on 14 occasions where 5 or more searches arrived, most
were processed within the 5 minute timeframe.
However on 27 occasions during non busy periods, where 3 checks arrived
within a 5 minute period, they were processed in longer than 5 minutes.
Conclusion
When looking at the current usage, whether it be from total numbers of
transactions, to hourly rates, to peak hours, the same conclusion can be made
that the current existing capacity built for the pilot is being grossly under
utilised.
It also has to be remembered that these findings are from an interim report, and
more detailed analysis is likely to be forthcoming in the full report in July.
However, the information in this analysis does allow sound conclusions to be
drawn and provide valuable guidance to any future system scalability.
As a side issue it also allows us to help identify if further numbers of devices can
be used without any need to expand the central checking capacity.
Version 0.3
RESTRICTED
Page 13 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
Before the key issues are considered , this paper will look at the 3 broad options
that have been identified. These are set out below
7. NATIONAL SOLUTION PROCUREMENT OPTIONS
When considering the three options, any solution will have 3 main components.
Component 1
Fingerprint Matching Capability
Component 2
Transmission method
Component 3
Devices
These can be depicted thus:-
NATIONAL PROCUREMENT OPTION 1 – Keep same device and provider and expand
to all forces
NATIONAL PROCUREMENT OPTION 2 – Keep the Fingerprint Database with
IDENT1, and use devices provided by other suppliers
Version 0.3
RESTRICTED
Page 14 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
NATIONAL PROCUREMENT OPTION 3 – Fingerprint Matching service and devices
provided by independent companies. IDENT1 provides database with ‘raw’ fingerprint
data.
These 3 options are now examined in a little more detail.
Version 0.3
RESTRICTED
Page 15 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
7.1 NATIONAL PROCUREMENT OPTION 1 – Keep same device and provider and
expand to all forces
This is probably the simplest and easiest option to achieve, and one that would
lend it self to a quick implementation across the police service.
The option sees the devices being procured on a national scale from the provider
used for the pilot (Sagem via NG).
The matching capacity at ‘central’ (IDENT1), is beefed up to accommodate the
increased workload.
The Criminal Justice Extranet (CJX) connection between IDENT1 and the
communications infrastructure is similarly revised to ensure it meets capacity
requirements.
Communications provided by 3G / GPRS and Airwave (when available) GPRS sim
cards should be provided and managed centrally, to obviate the connectivity
problems experienced in pilot.
Arguments for option:-
Tried and tested.
Delivers the basic requirements.
Easily implemented.
Training and support have already been put into place for the pilot, and
proven to be sufficient and effective. This can easily be expanded into the
additional 41 forces.
Extra devices would just be ordered and delivered to forces.
Additional service integration is simpler as there is only one device to
integrate with.
Device and software upgrades simple to implement
Arguments against option:-
Ties solution to one specific provider and one specific device
Device may not be suitable for all policing applications
May not meet ‘best value’ requirements of OGC reviews
May not be achievable under ‘single tender’ option.
May stifle research and development in external device manufacturers.
The solution is one that could be implemented very quickly but would not
necessarily represent best value.
Version 0.3
RESTRICTED
Page 16 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
COST
‘Central’ Upgrade to accommodate 15,000 devices
9 ‘stacks’ @ £260k = £2.34 M
Device costs – 15,000 @ £2,500 = £37.5 M (shared by
forces?)
Training :- Possibly provided by NPIA via the old Centrex
set up, and NG technical support.
Increase Help desk staff at NG
TIME FRAME TO IMPLEMENT
Mobilisation: 4 months
Device provision: 3 months lead time.
Training & Deployment:
From pilot – training for IT, users (train the trainers &
cascade), and go live can be achieved in one week per
force.
Minimum 14 weeks based on 3 forces per week (42 forces)
(This would be dependant on the available training
resources and NG engineers availability to commission
devices and train IT service desks)
Contingency: +15%
Version 0.3
RESTRICTED
Page 17 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
7.2 NATIONAL PROCUREMENT OPTION 2 – Keep the Fingerprint Database with
IDENT1, and use devices provided by other suppliers
This option would see Northrop Grumman maintain the fingerprint database, and
provide the fingerprint matching capability within IDENT1.
The interface that has been provided for the pilot, is capable of accepting
fingerprint encodings (also known as templates), as well as, or instead of, the
actual images (as provided by the pilot devices).
This option would allow other device providers, such as Identix, Motorola, and
Crossmatch, to bid and provide devices to the forces.
It is envisaged that this option would see a centrally provided system made
available for use. If forces wish to then take advantage of the system, they
would purchase suitable devices to connect to it.
The important part of this option would be the accreditation and certification of a
number of devices from different manufacturers. This would guarantee to forces
that if a particular device is shown on a ‘catalogue’ (similar to GCAT) that they
would have confidence that the device will work, and be sufficiently accurate.
Market forces would determine the types of devices that are available for
connection, and they could range from simple fingerprint only devices, to more
complex multi functional devices with a fingerprint capability built in.
A catalogue system would enable a centralised procurement framework to take
place, with forces being able to be ‘call off’ their requirements against the
framework, rather than have to go through the procurement process on an
individual basis.
It is essential that a simple system of accuracy checking is established to enable
current and new devices to be checked, assessed and authorised.
Northrop Grumman would be required to build the central architecture to
accommodate the expected usage, and to maintain response times agreed within
a national service level agreement.
Device type and communication type would be a matter of choice for individual
forces.
Arguments for:-
Utilises existing infrastructure which is tried and tested.
Allows for easier integration of interoperability options using the
fingerprint database at later date.
Version 0.3
RESTRICTED
Page 18 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
Depending on costing option, spreads costs away from central provision.
Competition for device provision may drive costs down to forces.
Complex multi functional devices could be introduced at any time
Larger choice of different types of devices for different scenarios.
Arguments against:-
No competitive costs for central matching – tied to existing provider.
Not conducive to quick implementation, however this could be offset by an
interim solution (see section 9).
Device availability, is subject to the individual forces ability to fund. May
create a ‘have’ and ‘have not’ situation.
Multiple vendors introduces additional risk and complexity
COST
‘Central’ Upgrade to accommodate 15,000 devices
9 ‘stacks’ @ £260k = £2.34 M
Device costs – up to 15,000 @ £1,500* = £22.5 M (shared
by forces?)
* average price per device.
Training – The training will be device specific, coupled with
some knowledge of how the central system works, along
with results interpretation.
TIME FRAME TO IMPLEMENT
Mobilisation
4
months
OJEU & ITT
6-12 months
Selection, testing, and authorisation of appropriate devices
3-6 months initially. Ongoing authorisation process to be
developed to accommodate new devices coming onto the
market.
Procurement & delivery – 3 to 6 months
Training and deployment – 3 to 6 months depending on
method used.
Contingency: +15%
Due to the relatively long timescales to implement this option, an interim
solution could be put into place. For example the current pilot could be
Version 0.3
RESTRICTED
Page 19 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
enhanced and a limited capability introduced to all forces in the interim period
(see section 9).
Version 0.3
RESTRICTED
Page 20 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
7.3 NATIONAL PROCUREMENT OPTION 3 – Fingerprint Matching service and
devices provided by independent companies. IDENT1 provides database with
‘raw’ fingerprint data.
The essential part of this option sees the matching function being completely
separated from IDENT1.
The fingerprint data held within IDENT1, is wholly owned by the police service.
This option will see the fingerprint data being exported on a regular basis from
IDENT1 to a separate 3rd party system (but still within the police service).
This option can offer a variety of possible solutions.
For instance:-
It could be that a sole provider offers a complete ‘end to end’ solution from the
matchers to the capture device, including various communications methods.
It could also be that one provider supplies the matching capability, while a
myriad of suppliers provide the devices.
Arguments for:-
Gives option to design system from top to bottom.
Does not tie police service to existing supplier.
Competition for contract should deliver good value for money.
Keeps service independent from other fingerprint work.
Complete flexibility to enable any approach to be considered.
May be able to host additional services such as FIND
Arguments against:-
May cause problems for future integration and enhancement work.
Database needs to be updated on regular basis.
Could take considerable time to implement
Still reliant upon connectivity to IDENT1
In 2006 the project team took the opportunity to use an external company to
provide a document giving indicative costs to set up from scratch a system
identical to that of the current ‘Lantern’ system.
It includes the setting up of the database, importing the data from an external
source (IDENT1), maintenance of the database, providing the devices, and
licensing the officers.
The costs and timescales are lifted directly from the document provided by the
company New Zealand Biometrics. This company has since been bought out by
other suppliers of fingerprint systems.
Version 0.3
RESTRICTED
Page 21 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
COST:-
100 node matcher system providing end to end response
time of 4.7 seconds
£315,000
20,000 devices @ £565 per device to include all software
licences
£11,291,000
20% spares £1,388,000
Software and licences inc upgrades
£700,000
Server application one off cost for 8 million records
£588,000
Application support annual fee
£117,000
Cost of importing 8 million records
£762,000
Annual support costs (20%)
£160,000
Police officer enrolment (100,000)
£100,000
Total cost of end to end system (inc devices)
£15,421,000
Total cost - Excluding device costs
£4,130,000
TIME FRAME TO IMPLEMENT :-
Between 18 to 22 months
RECOMMENDATION 2
It is recommended that
National Procurement Option 2 is
selected as the most appropriate method to take the project
forward.
Version 0.3
RESTRICTED
Page 22 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
8. PRINCIPAL KEY ISSUES OF ANY NATIONAL SOLUTION
There are a number of key issues that can be considered to be principal in the
decision making process of any national solution, as they will most definitely
dictate the direction the solution can take, and the time and cost involved in
delivering that solution.
8.1 PRINCIPAL KEY ISSUE 1 - DEVICE SIZE
One clear issue to be considered is the actual device size. The device used for
the pilot, is of a size and weight that does not lend it itself to be easily carried by
officers on foot patrol duty.
The users have asked for consideration of the device to be made smaller, or to
have the fingerprint functionality incorporated into a device already used by
them on patrol (such as PDA’s, Blackberry, etc)
In the device being used for the pilot, a single finger, optical fingerprint reader is
used. This works on the principle of reflected light through a prism. The optical
prismatics govern the overall size of the fingerprint device.
E.g.
A single fingerprint reader of this type is the size of a computer ‘mouse’.
A two finger reader is typically the size of a house brick,
And a four finger device the size of a briefcase.
This type of fingerprint reader has been chosen for its accuracy in checks
involving what is termed ‘1 to many’ ( 1 ~ n ).
Other types of fingerprint reader do exist. These can best be described as
capacitive, optical bar, and flat optical. Although the methods used for
extracting the fingerprint information differ between the devices, their common
feature is that they are all thin and flat, so could easily be incorporated into a
device such as a PDA.
These applications are widely used in private industry for security type
applications for verification of identity, and have been found to be very effective.
However the types of check that are undertaken are invariably termed as ‘1 to
1’. The device is essentially comparing a print offered by a subject against a set
of prints previously enrolled against the subject name, or other identification
token such as a smart card. It is answering the question “ Is this the finger print
of John Jones?”, rather than the question, “ who does this fingerprint belong to?”
Version 0.3
RESTRICTED
Page 23 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
Where checks are made against a database, where the system is not directed to
look at one particular set of prints ( as described above) ( a 1 ~ n check), the
databases are usually small, and can involve a high false reject rate (FRR)(i.e.
the database doesn’t identify the individual).
This type of ‘non identification’ usually only means that the individual cannot
gain access to the function they want to do, such as to enter a building, or log
onto a computer, so does not present a particular huge problem apart from
inconvenience.
The subject will usually have alternative methods to deal with their non
identification, because they themselves want it resolved.
In the police environment non identification would have a much greater impact,
as it effectively would mean a known criminal not being identified. (It is doubtful
the criminal would want to seek alternative identification themselves. )
Along with a higher false reject rate, the devices typically also produce a higher
‘false accept’ rate (FAR) (wrongly identifying individuals as being on the
database when they are not).
In a police environment this type of ‘wrong’ identification, has a potential to
undermine the use of the device, as well as opening up the police service to
potential civil proceedings for unlawful arrests. Although with any system that
involves machine checking against machine, such misidentifications will occur, it
is necessary however to keep them to very low levels.
The reason for these high FRR and FAR rates in capacitive or thin membrane
readers, is probably due to the fact the reader samples a substantially smaller
fingerprint area than the prismatic optical reader.
The industry leaders, current research, and the NPIA biometrics experts
agree that these types of fingerprint reader are, as yet, not of an
acceptable quality for the purpose to which we wish to use them.
It is important to note that the FBI only approve such devices for ‘PIV’
(personal identification verification). They still advocate optical devices
for forensic quality examination.
To be able to use such readers, it may be necessary to stimulate the market,
and research in these types of devices, to get the improved accuracy needed.
This approach would not lend itself to an early implementation.
Capacitive type devices have been used in Iraq and Afghanistan to good effect
however no information exists in the public realm as to what the systems consist
of, FAR & FRR rates, etc.
Version 0.3
RESTRICTED
Page 24 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
Some claims of accuracy by manufacturers of capacitive devices are impressive,
however it usually has been achieved by a proprietary end to end system
including feature extraction algorithms that compliment the central encoded
files.
Such capacitive or flat optical solutions cannot be completely discounted, but it
would be up to manufacturers to adequately demonstrate that the devices they
offer give sufficient accuracy with acceptable FAR / FRR rates.
RECOMMENDATION 3
In preparing system requirements for the tendering process, only
accuracy, and associated False Reject / Accept rates should be
specified. The burden will then be put upon manufacturers to
demonstrate how their device meets that requirement. This should
allow a broad spectrum of devices to be tested and approved.
8.2 PRINCIPAL KEY ISSUE 2 - ACCURACY
The accuracy tests conducted with the device currently being used on the pilot
have shown a high degree of accuracy, typically 94 -96%. Although not as high
as Livescan (99.98%), it is sufficiently high enough for police purposes as a
‘screening’ device on the street.
The original accuracy tests were conducted with a relatively small reference set
(between 400 and 500 police eliminations), but these were ‘launched’ against a
database of over 6.5 million. Operational use of the device has certainly
indicated that the accuracy rate is probably higher than that achieved in the pre-
pilot tests.
The device has also had a very low false accept rate (about 0.6 ~ 0.7%).
These accuracy rates have been achieved using only two ‘flat’ impressions of the
subjects index fingers.
It has been commented upon that the accuracy rate of the device could be
improved if more than 2 fingers were used in the checks.
It is true that the more fingers used in the comparison, the higher the accuracy.
It also reduces the amount of processing power needed when carrying out the
checks. This is perhaps counter-intuitive but non the less is correct.
The reason the pilot used only two fingers, was to reduce the opportunities for
police officers to make mistakes in the correct sequence of fingers.
Version 0.3
RESTRICTED
Page 25 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
The central fingerprinting architecture is such that it relies on the prints being
presented in a particular order and in a particular orientation.
Livescan, uses the two sets of 4 flat impressions taken first to determine the
correct finger sequence of the rolled prints. If an officer attempts to take the
prints in the wrong order, the Livescan device will detect the error and display it.
No such ability exists with the current lantern device. It will detect if the same
finger is taken twice, but will not detect if the officer has taken left then right,
instead of right then left. The result of the reversed sequence is that if the
subject is on the database, they will not be identified.
By introducing more fingers into the process, it introduces the higher probability
of the fingers being taken out of sequence. It increases the time the process
takes on the street, and also increases the resulting file size to be transmitted
(see communications later).
There is a direct trade off to be achieved between the numbers of fingers, the
accuracy of the check, and the simplicity of the process at street level.
RECOMMENDATION 4
In preparing system specifications, it is recommended that the
simple two process operation for obtaining fingerprints is retained.
This should not preclude multi finger devices from being put
forward for approval if they can demonstrate that they meet the
specifications.
RECOMMENDATION 5
It is further recommended that a system of accuracy testing is
developed as a matter of urgency, to enable easy assessment of
devices for approval.
Version 0.3
RESTRICTED
Page 26 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
8.3 PRINCIPAL KEY ISSUE 3 - IMAGES AND ENCODINGS (TEMPLATES)
This subject has an impact on several areas of the project, and contributes to the
variability in the options to be considered.
The principle difference between the two, as far as the project is concerned, is
that where an image of the fingerprint is captured, the whole image (which may
be compressed, but none the less ‘raw’) is then sent to the central processing
system, where the print is then encoded , and searched.
Encodings or fingerprint template, on the other hand, result from the process
that the device first captures the image of the fingerprint. The device then
processes the print on the device to extract the minutiae, from which an
encoding or template is obtained. The resultant file is then sent to the central
processing system to be searched.
The following table summarises the differences.
Images Encodings
/Templates
High quality
Quality dependant on device algorithm
Encoding done at ‘central’
Encoding done at device
High processing power needed at
Less processing power at central
central searching facility.
Device becomes searching algorithm
More processing power needed on
independent
device
May drain battery quicker because of
extra processing needed.
Large file sizes to be transmitted
Smaller files to be transmitted
‘central’ has the matcher licences
Individual Devices need licensing
Accuracy can be virtually guaranteed
If device algorithm not the same as the
as the encoding and searching is done
central matching system, additional
with the same algorithm.
translation software will be needed.
May have reduced accuracy.
The key part to this issue is the file size of the fingerprint(s) that
needs to be transmitted. This has a direct impact on the
communications system that can be used to transmit it from the
device for processing. It also impacts on where the processing
takes place, the processing power required, and the licensing
arrangements.
Version 0.3
RESTRICTED
Page 27 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
The IDENT1 interface that has been constructed for the pilot, is capable of
receiving and using encodings, but as yet this functionality remains
untested.
The use of images means that the device being used to take the
fingerprint is not tied to a particular algorithm. This would allow multiple
manufactures to supply devices, provided they could meet some minor
requirements.
The use of encodings may not be a barrier to allowing multiple vendors.
The reason for this being is that essentially there are only 3 or 4 encoding
algorithms. Thus the manufacturers are now building their devices to be
‘algorithm agnostic’. This meaning that the device manufacturers will load
whatever algorithm is to be licensed. Essentially the same algorithm as
used in IDENT1 (Sagem) could be licensed for each device. This should
help keep accuracy levels high.
Where the device provider cannot use the same algorithm, conversion
software does exist. However there is an impact on accuracy which would
need to be assessed.
Using encodings essentially will also future proof the system, and will
allow non optical readers to be used provided they are able to meet the
accuracy requirements.
RECOMMENDATION 6
It is recommended that the IDENT1 interface continues to enable
the use of encodings and images in any future solution. This will
allow multiple vendors of devices to be considered, and future
proof the solution.
Version 0.3
RESTRICTED
Page 28 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
8.4 PRINCIPAL KEY ISSUE 4 - COMMUNICATION METHODS
For the pilot, the method of communication being used is GPRS.
This method was decided upon due to the fact that the file sizes being
sent were relatively large (20Kb), and that at the time the pilot was
commenced AIRWAVE, the police tetra based system, was not able to
prove that it could handle the file sizes in an acceptable time. Its poor
data rate (less than 4Kb) and ‘single slot’ for data, in its transmission
spectrum, suggested that this would induce a significant delay in
transmission of the fingerprint, and could impact significantly on the
AIRWAVE system.
Initially the project was looking at a vehicle based system, which may
have been able to take advantage of the AIRWAVE system as the
transmission medium, assuming the system was able to transmit the
fingerprint files.
However, a technological improvement in communications with Personal
Digital Assistant (PDA) type devices, meant that a self contained device
could be used, which better suited the police requirements. It also
represented a huge leap forward technologically allowing the project to
remove an intermediate ‘step’ in any final solution.
The vehicle based solution was thus abandoned in favour of a self
contained solution. This decision also precluded AIRWAVE from being
used for the pilot, as all the self contained, mobile fingerprinting devices
considered suitable for the pilot, are Personal Digital Assistant (PDA)
based.
At the commencement of the pilot there was no PDA available that was
tetra / airwave enabled, that could be used in the fingerprinting devices.
At the present time there is still no commercially available PDA that is
AIRWAVE enabled. ( see below re experimental device).
AIRWAVE, since the pilot has been in operation, has made moves to try
and improve its data handling capabilities, and promises to be able to
handle larger size data files.
The project team have worked closely with the AIRWAVE / mobile data
team to examine the viability of AIRWAVE as the data carrier.
Version 0.3
RESTRICTED
Page 29 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
An AIRWAVE enabled prototype PDA is now available and tests have been
recently conducted to assess its suitability. These tests, using multi slot
data capability on the Airwave system, have shown that it is possible to
use the Airwave system to send fingerprint images to IDENT1, and
receive responses in time frames comparable to that achieved by GPRS.
The response times could be further improved if only encodings are used
as the resulting file sizes are much smaller and will take less time to
transmit.
However, although Airwave has now been shown to be able to cope with
the large file sizes associated with fingerprint images, the problem still
exists that a commercially available, Airwave enabled PDA does not exist.
This may be resolved in the long term if PDA manufacturers find it
financially viable to produce them.
In the short term this may severely hamper mobile fingerprint
manufacturers in providing an Airwave enabled device. This is because
the mobile fingerprinting devices almost without exception, are
commercial PDA based.
Thus if AIRWAVE is able to provide the communications, there
may not be a device capable of taking advantage of that method
either in the short or medium term.
The availability of devices aside, both Airwave and GPRS/3G have their
advantages and disadvantages, which can also impact upon the decisions
made by forces on the devices that they will need.
These are summarised in the table below.
AIRWAVE
GPRS / 3G
At present cannot handle large file
Can handle large file sizes expediently
sizes expediently (- )
(+)
Has secure communications (+)
Needs additional security programmes
to protect transmissions (-)
PDA not generally available to device
Uses COTS types of PDA, and
manufacturers (-)
communications programmes. (+)
Guaranteed to be available in times of
May not be available in times of crisis
crisis, as it uses own system. (+)
as it uses public networks (-)
Restricted to use within the UK (-)
Can be used anywhere as long as PDA
communications are tri / quad band
and SIM is enabled for roaming. (+)
Coverage is as good as each force’s
Coverage can be a bit ‘patchy’, and
communications coverage. (+)
dependant upon 3rd party provider. (-)
Version 0.3
RESTRICTED
Page 30 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
It is interesting to note that current mobile data trials using PDA’s has
chosen to use GPRS as the transmission medium rather than Airwave.
However this may be directly related to the previous observation made
about the dearth of commercially available Airwave enabled devices.
RECOMMENDATION 6
In setting system specifications, it will be necessary to ensure that
both 3G / GPRS and Airwave can be accommodated, (not
necessarily in the same device), so that forces can choose to have
either system.
For Disaster Victim Identification, and other requirements such as
travelling football fan identification abroad it is imperative that 3G/
GPRS is used.
Version 0.3
RESTRICTED
Page 31 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
8.5 PRINCIPAL KEY ISSUE 5 – FUNDING
Funding will inevitably be the major factor in any proposed roll out to all forces.
Currently the project cover the costs of providing the devices on lease from Northrop
Grumman, along with service and support costs.
The 10 police forces engaged in the pilot, provide the first level of support for the
devices, as well as paying for the cost of the GPRS data transmissions according to
their own arrangements with the service providers. Thus costs to forces are minimal.
This may not continue to be the case when considering a national system.
Many forces may wish to own their own devices, as part of their own asset
management, rather than leasing them.
One device type may not meet all the needs for the various deployments within the
police service, so a catalogue of devices may need to be developed according to the
types of deployment they are being used for.
Depending on what level of cost is involved, the take up of the service by forces may
be restricted, despite the obvious advantages that having such a device may bring.
Thus serious consideration needs to be given as to the types of funding methods
that can be used.
The methods can be summarised as being.
Fully funded from the ‘centre’ (home office), as a centrally provided service as
like the PNC, being ‘top sliced’ from overall police funding
Shared funding between home office and the forces, in a manner similar to
Livescan.
Costs fully borne by the forces.
With force budgets continuing to be squeezed, it is hardly surprising to note that the
preferred option of the forces is for the solution to be centrally funded.
Unfortunately pressure on central government budgets is equally as strong and thus
may not be able to provide the required level of funding.
Some forces both inside and outside the pilot, have voiced the fact that they would
be willing to find funds to purchase the devices / or service but it would dependant
upon the costs.
Version 0.3
RESTRICTED
Page 32 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
Clearly larger forces may have the ability to find sufficient funds to finance
operationally viable numbers of devices more easier than smaller forces. The
benefits identified to policing are such that any restrictions on deployment because
of the lack of funds should be removed.
One way of achieving this could be to allow forces to offset expenditure on Lantern
devices / services against the annual savings and efficiency requirements.
It is important that any decisions on funding methods are made early enough to allow
forces to include any funding requirements in their 2008/9 budgets.
RECOMMENDATION 7
Once the option to go forward has been agreed, urgent talks are
had with NPIA senior executives, and the Home Office to agree
funding method.
This agreement should be reached in sufficient time to allow all
forces to consider their requirements for 2008/9 budgets. (this
usually takes place in late September / early October)
Version 0.3
RESTRICTED
Page 33 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
9. PILOT EXTENSION - Opportunities to exploit under usage
The usage analysis highlights an under usage of the current capacity. This under
usage could be exploited to provide a modest expansion of the pilot as an
interim to the final national solution becoming available.
The justification for an extension of the pilot, can be drawn from the huge
interest that is forthcoming from forces outside the pilot, as well as the pilot
forces themselves wishing to expand their areas of operation and device
availability.
The volume of success stories coming back from forces, and the benefits that are
already being delivered in measurable quantities, makes it almost seem criminal
to restrict those benefits to a small number of forces and users.
It must be understood that a fine line has to be walked when considering any
expansion, to avoid any ‘de-facto’ solution being given to the service, that stifles
future development and opportunities to keep costs down through competitive
market forces.
It is with all that in mind, that the project board are asked to consider
the following options that are available to exploit the current under
usage of the existing pilot checking capacity.
Expansion Option A - Keep the 10 forces and expand the number
of devices in each force
Expansion Option B - Expand the number of forces, but limit
numbers to 10 (or so) per force
Before any consideration can be made for an expansion of the current pilot, it is
necessary to identify to what extent can the current system be expanded, and
what risks are associated with any decision on numbers.
Version 0.3
RESTRICTED
Page 34 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
The following graph illustrates what Northrop Grumman believe to be the impact
on device numbers against central matching capacity. As can be seen the graph
shows that the current 200 searches per hour can support up to 300 devices
without the need to expand the matching capacity in ‘central’.
If 600 devices were to be deployed, a central matching capacity of 400 searches
per hour would be needed, and a capacity of 700 searches per hour could
support around 1000 devices.
1400
r
.)
1200
e
s
/
H
h
Less frequent search
1000
responses over 5 minutes
e
a
rc
(S
800
a
c
i
ty
a
p
600
e
r
C
More frequent search
responses over 5 minutes
t
c
h
400
t
r
a
l
Ma
200
e
n
C
0
0
300
600
900
1200
1500
1800
2100
# MFRs
These figures should be seen as the ‘safe bet’ to achieve the response times of
less than 5 minutes.
It needs to be understood that more than 300 devices could be deployed with
the central matching capacity remaining at 200 searches per hour, but to do so,
would increase proportionally the likelihood of more instances where a 5 minute
response would be exceeded.
The capacity argument above also assumes that the current pattern of use
(mainly week days during the day) is continued in any further deployments.
If the further deployments were purely in support of ANPR this may be the case,
However if the expansion were used to support general policing activities, this
Version 0.3
RESTRICTED
Page 35 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
could perhaps see the use being spread more evenly, or certainly wider across
the 24 hour / 7 days a week
The graph below shows that a theoretical extrapolation of the peak transactions
per hour currently achieved in the pilot (22) can be used to calculate the possible
number of devices that a capacity of 200 checks per hour could support.
No of MFR’s served by Central capacity of 200 searches per hour
Thus it can be seen that a number between 300 and 900 devices could be
deployed using the existing capacity.
The risk identified being that the larger the number of devices over 300, the
increased likelihood that search times on occasions of peak usage would exceed
the 5 minute response parameter set for the pilot.
One further thing to be borne in mind is that in the current transactions, the vast
majority of the overall transaction time is taken up by the encryption,
transmission, and decryption of the fingerprint and result data, rather than the
time spent on the matchers.
Version 0.3
RESTRICTED
Page 36 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
Expansion Option A - Keep to the 10 pilot forces and expand the
number of devices in each force
This is the simplest option to expand the pilot. Training and support is already in
place, and all it would require is the supply and commissioning of additional
numbers of devices.
This option would be wise to limit the numbers of devices to the 300 identified by
Northrop Grumman as being the optimum number of devices that can be
supported by the current capacity limitations (200/hr).
300 devices would give an average of 30 devices per force. Forces currently have
on average 10, so this would see an additional 200 devices being deployed, with
a commensurate number of spares being made available ( currently 10%)
It would be advisable to allocate these additional devices in proportion to force
establishments.
COST:-
200 devices @ £2,500 = £500,000
300 devices @ £2,500 = £750,000
10% spares 20 = £50,000 30 = £75,000
TIME FRAME TO IMPLEMENT :-
dependant on manufacturer
Believed approximately < 3 months from time order
placed.
Hurdles/ Items to consider :-
1. Funding – Who pays for the devices
2. Will the devices still be subject to lease as per the first 100
3. Current device is obsolete, will the original 100 be replaced with ‘new’
device.
4. GPRS or Airwave?
5. CJX capacity
6. Certificate Authority – capacity, servers, hosting
Version 0.3
RESTRICTED
Page 37 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
Expansion Option B - Expand the number of forces, but limit
numbers to 10 (or so) per force
This would see the remainder of the forces (41) in England, Scotland and Wales
getting a number of devices each, similar to the numbers given to the pilot
forces.
If a limit of 300 devices is imposed this would see a distribution of 5 devices on
average to the remaining forces. This number would probably not be worthwhile
deploying in some of the larger or more rural forces.
If an average of 10 were deployed to each force in line with the current pilot
forces, this would take the total number of devices to 500 (including the 100
already deployed).
This number, if the report by Northrop Grumman is taken into account, could
introduce a significant risk that the frequency of times the checks exceeded 5
minutes.
COST:-
400 @ £2,500 = £1M
500 @ £2,500 = £1.25M
Spares 50@ 2,500 = £125,000
TIME FRAME TO IMPLEMENT :-
Device provision :-< 3 months lead time .
Training & Deployment :- 14 weeks based on 3 forces per
week (dependant on training resources and NG engineers)
Hurdles/ Items to consider :-
1. Funding – Who pays for the devices
2. Will the devices still be subject to lease as per the first 100
3. Current device is obsolete, will the original 100 be replaced with ‘new’
device’.
4. GPRS or Airwave?
5. CJX capacity
6. Certificate Authority – capacity, servers, hosting
Version 0.3
RESTRICTED
Page 38 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
RECOMMENDATION 8
It is recommended that Expansion option B is undertaken if
expansion to the pilot is agreed. This will enable a limited
capability to all forces while the final solution is developed
Version 0.3
RESTRICTED
Page 39 of 41
RESTRICTED
Lantern
Project options for pilot extension and national capability
Appendix A - Costing assumptions
1. Devices chosen are only available to police officers on duty
2. PCSO’s are excluded from those persons requiring use of the device as the
dormant legislation restricts use to Police Officers only.
3. Devices are only available to the 140,000 police officers of the rank
Constable to Chief Inspector
4. Shift patterns adopted by UK police forces allow a maximum of 1/3 of
establishments to be on duty at any one time (approx 46,000)
5. Of the 46,000 approximately 1/3 are unavailable due to sickness, courses,
and leave. (leaving 30,000)
6. Devices would not be issued on a 1 to 1 basis, more probably a 2 to 1
ratio. (15,000)
7. Even if the capability were provided on an individual basis, it is unlikely
the system would see more than 30,000 concurrent users.
8. All costings are based on the figures provided by NG in their CCN proposal
for the Lantern Pilot, unless specifically attributed elsewhere.
9. Device cost for pilot was estimated at £2,500 each. It is assumed that the
cost included hardware, and all software licensing.
10. ‘Central’ upgrade to handle 15,000 devices. Each ‘stack’ of processors can
handle up to 1800 MFR’s therefore 15,000 will require 9 stacks.
11. Each stack has 15 banks of matchers. The Pilot only had one stack with 4
banks of matchers and cost £259,600
Version 0.3
RESTRICTED
Page 40 of 41
Appendix B - Decision making tree – Lantern Options
RESTRICTED
