UK Immigration Service's Suspect Index Project

By Dom Pancucci

Every nation wants to ensure that criminals, terrorists and other undesirable people do not cross its frontiers. The international nature of much criminal activity and an unprecedented freedom of movement for many people makes this task increasingly difficult, with the greatest problem being the maintenance of accurate and timely data on suspects at the points of entry to a country. Over 40 million people visit the UK in a year. Yet the UK Immigration Service reckons it has found more than a partial solution to the conundrum of providing useful data to its officers at airports and other disembarkation points. And the twin goals of immigration control are not only to spot dangerous people of various kinds, but to also to make it as easy as possible for legitimate visitors to gain entry.

The system chosen for the Immigration Service's so-called Suspect Index (SI) is based around Unix and the Oracle database running on RISC-based servers from ICL, with clients being either desktop PCs or laptops. SI also includes an object-oriented element founded on DAIS, ICL's request broker based on CORBA, the standard developed by the Object Management Group. Called PowerVision, the sub-system powered by CORBA is used for supplying images of people held on the SI system to immigration officers. The project was awarded to UK vendor ICL, now owned by Fujitsu, which effectively supplied a turnkey solution.

SI's project value was 23 million (pounds sterling), or around $34 million. The SI system can be used to screen anyone seeking entry to the UK who does not carry a passport issued in a European Community state or Scandinavian countries involved in the EFTA trade alliance. The curious thing is that an Australian seeking to enter the UK will be checked, while a Swedish person will be able to go through the fast channel.

Previously the organization had looked at a number of ways to get data to its officers, according to Ken Richardson, assistant director at the Immigration Service Enforcement Directorate and project manager for SI. None proved to suitable for widespread deployment, however, before the SI system was put together. Printed lists which were guaranteed to be current were the default option.

"Suspect Index dates back to studies done in the 1970s, where we were looking for an application which could perform the task. There were a number of false starts, usually because the technology was not available. So by the 1980s we decided to develop our own prototype system," explains Richardson.

As a result of this decision to develop in-house a database of suspects, the Immigration Service came up with a system based on DEC's MicroVAX platform. This held a database which allowed officers to scroll through for names related to a passport or other suspicious sign. Initially the system looked successful, so two immigration desks at the Dover sea port on England's south coast were set up with it on a trial basis. Once in use at a live location, however, it became rapidly clear that the endless scrolling required to search the database lists was too slow and inflexible. So the project was canned.

After the disappointment of not being able to deploy its initial attempts at developing a system, the Immigration Service started a truly in-depth analysis of all its real requirements and future goals for creating an effective index of suspects, according to Richardson. Surveys were conducted among both management, end-users and the data suppliers over a three month period about what should constitute the best possible service. From these responses the description of what the SI system should become boiled down to a number of key features.

"The system we wanted had to be as reliable as a book, as big as a computer system and easy to use, with 24-hour availability and upgrading of information," says Richardson.

Once the basic requirements of the system were established, Richardson and his colleagues then had to develop a business case for creating the system.

Once the basic requirements of the system were established, Richardson and his colleagues then had to develop a business case for creating the system. In this context, areas like social security payments were looked at. Questions were asked. What would be the cost to the UK government if the wrong people were let in and claimed benefits? Similar speculation was conducted about a whole range of issues related to the state and what costs and disadvantages would ensue from lapses in immigration control due to a lack of accurate and timely. Richardson cannot reveal the estimates which were arrived at by the Immigration Service and other UK government departments, but he insists that some of the predicted benefits have come through now that the SI is up and running.

Following the clear case for developing SI, the Immigration Service put the project out to tender. Three companies were on the short list, but ICL won the day. ICL got the vote because of the quality of its design study, its clear interpretation of the Service's data needs for the SI project and its competitiveness on price. Work on SI started in January 1992. Right from the word go, however, Immigration Service staff and ICL people sat down and hammered out all the necessary requirements. Conflict resolution protocols were agreed upon from the outset, to cover any major disagreement which might arise during the collaboration. Eighteen months previously, a system had been installed at the Immigration Service which did not meet operational requirements. While not personally involved in this affair, Richardson clearly drew lessons from what happened.

"It was vital for us to all have a common view and it was stated clearly that no system should go in before all potential faults were known as close as this can be realised. It was then up to us to make a decision about going live with the system, based on the known faults," Richardson says.

The more technical side of the system's development was left largely to ICL. The Immigration Service did not specify what types of technology should be incorporated into SI, with ICL given the freedom to choose the system elements based on the agreed functional and operational requirements. A pilot system was installed at Stansted airport early in 1995, with most of the faults ironed out, leaving only 11 known problems left after the system had been "thrashed for 16 hours a day" in Richardson's words. For instance, under certain circumstances a bug emerged in the Oracle 7 database used to house the SI system's information store, which Oracle worked hard with ICL and the Immigration Service to solve.

A further series of operational tests were run on SI, before it was decided to deploy the SI system at all major entry points within three months. Locations included the major airports such as Heathrow and Gatwick, as well as the Channel Tunnel and major sea ports, such as Dover. In total 23 installations took place within the first tranche. Yet deployment turned up its own problems, mainly due to logistics.

"It is difficult to put systems into live immigration control situations, due to the high numbers of people coming off flights from places like the US and the Far East. We had to install the system in little bits and run four hour training sessions as terminals were actually being installed," Richardson recalls.

During these "live" training sessions, it turned out that people working intensively with the SI terminals did not want to give them up. And as user input was highly valued during the trials, the immigration officers felt a greater sense of ownership of the project, as they could see there ideas captured within the GUI and other features. Because of this and other reasons, Richardson reckons that there was a collective will to see the SI system work out. There are now around 75 operational SI sites at any one time, according to Richardson, with around 3,000 users of the system in total. By December 1996 all initial phases of the project will have been completed, such as installation and training.

SI is divided into three separate sections on the terminal screen. Immigration officers can search the data store by people, document and organisation. While the first two categories seem logical choices, there is also apparently the need for certain suspect organisations to be considered as worth an alert at the point of entry. Certain people trying to get into the UK may appear to have valid documents, yet name a bogus or defunct company, English language school or other organisation as the reason for the visit. In these circumstances, further questioning of the visitor would be required to meet entry requirements.

The user interface remains identical for all three sections of the SI system. The keyboard of the client terminals used for the project were also customised, with optical character recognition systems installed equipped with a swipe reader for passports. A full Windows GUI has been developed for the back office computers. Where there are temporary immigration control points, or vulnerable equipment is left unattended for long periods, officers are equipped with notebook computers supplied by a vendor called Twinhead.

Over 700 terminals have now been installed running SI. System response times for data retrieval has been brought down to under a second, while downloading an image to the SI workstations takes only between eight and 12 seconds.

Several copies of the main SI database are held on servers located at the point of entry. The central Oracle database has over 400,000 entries on suspect individuals, including possible aliases. Over 1 million listing could be on the system by the year 2000. Currently SI holds more than 40 times the number of names held on the previous paper based method of checking for suspect visitors. With images the main Oracle database reaches 30 GB in size. Over 700 terminals have now been installed running SI. System response times for data retrieval has been brought down to under a second, while downloading an image to the SI workstations takes only between eight and 12 seconds.

The central database engine is held on two ICL DRS6000 machines, running standard Unix System V. There are also around 90 ICL TeamServers scattered around the UK, holding the mirrored databases essential to maintain service resilience. This was another factor Richardson and his colleagues discovered during the user evaluation phase - that no technology was available to maintain speedy and high availability access to a database held at some central site. The whole system is protected by security features complying with ITSEC E2 levels.

Using the CORBA-based DAIS backplane was the best solution for handling the imaging side of the system, according to Trevor Hoath, a principal consultant at ICL and part of the original SI development team. ICL had to deal with supplying images to over 100 locations in total for the SI project. This decision was made, however, at a time when object technology was still considered a youthful science. Yet Hoath maintains that only through using DAIS could the need for fast access to images be satisfied. Two capabilities had to be fulfilled regarding the imaging side of SI.

"Under SI, images have to distributed very easily from a central point to multiple locations and the system has to be scaleable without the need to change the application depending on the size of the image bank," Hoath observes.

Hoath reckons that only the ICL bid offered the Immigration Service the high performance necessary to meet the demanding tasks laid down by the SI project. From the officer's point of view at the immigration control booth fast access and ease of use are the essential prerequisites for such an information retrieval system. Waiting for up to 12 seconds for a high quality, large format image of a person to appear is also an acceptable time.

The whole data store for the SI system is updated daily generally, although urgent changes could be made more quickly. Current information is sent over the Ethernet network through using the X.400 messaging standard, from the central Unix servers to the TeamServer machines located at many of the Immigration Service sites. A WAN links all the sites together, with the central node being situated at Heathrow.

"The SI project is a classic example of using the right technology to fit the job, with a hybrid solution being the key feature. We achieved the best usage of IT, using object technology to handle the imaging. But the system would not have performed had we gone with a totally object-oriented system," says Hoath.

There is no need to link the SI database to other systems in use with the Immigration Service, according to Richardson, due to its highly specialised purpose. ICL and its parent company Fujitsu retain the intellectual property rights to the system, however, which means there is possibility that the technology package could be resold to third parties. This could only take place following consultation with the Immigration Office. Names matching within information systems like SI is a hard thing to achieve, so there may well be potential adopters of the basic networked database concept.

One possible future usage of the SI technology is for a similar pan-European system still at the planning and early development stage. A big central system may evolve from this work, which will need to interface with satellite installations. SI could easily be included in this plan, in particular due to its messaging engine being based on X.400, which is a readily available technology. At present this potential project is being called the European Information System.

Proof of the benefits available from the SI system can only be validated by a real world user. Richard Harlow is an immigration officer working at Heathrow airport - one of the busiest transit points in the world. In fact, Harlow is a controller working at the airport's Terminal Three and eventually had responsibility of training other officers in using the SI system, now acting as a mediator between the help desk and the users. Since SI was installed at Terminal Three in March 1995, Harlow and his colleagues have been using it for well over a year and a half.

"I was involved in the project from the start and spent a few weeks with ICL doing user testing and fault finding. Then I had to train others, but the system is very easy to pick up. People with no computer literacy could use it," says Harlow.

SI is a major leap forward compared to the earlier use of a book with suspects' names and details, according to Harlow. While the book was updated daily, its accuracy was reliant on officers putting in the relevant information, which was not a wholly satisfactory method. Now instant updates are available, which Harlow reckons has delivered great benefits.

"There are instances where someone is denounced as a suspect through a phone call, or letter with specific information on the person's travel plans, even down to a particular flight. This type of information is now available to us straight away and this was not possible before," Harlow observes.

Any technology used in immigration control remains merely a tool, however useful that tool turns out to be. Harlow and the other officers still have to make all the decisions based ultimately on their own judgement.

Any technology used in immigration control remains merely a tool, however useful that tool turns out to be. Harlow and the other officers still have to make all the decisions based ultimately on their own judgement. To this extent there will probably never be a system to replace this human factor. Yet Harlow confirms that the SI project has been a success. It provides the immigration officers with invaluable resources. Even the most stubborn person tends to end denial when confronted with a clear picture delivered from the suspect image bank. Greater efficiencies and time saving can be listed as tangible benefits, according to Harlow. There is also the added bonus that legitimate visitors to the UK can now get through immigration controls more quickly.

Richardson is not at liberty to disclose the kinds of saving the UK can expect from the deployment of SI, but the payback period in Immigration Service terms will only take around four or five years, after which the project will move from being a "debit to a credit", according to Richardson. And even though the system is now operational and proving its worth, Richardson's department is still polling the user base to seek ways to improve SI and sort out any lingering loose ends.

Integration at a Glance

The UK Immigration Service needed to solve the problem of getting timely and accurate information to its officers working at points of entry to the country. The solution was a hybrid system to gain maximum performance.



Back | Table of Contents | Next