Introduction The purpose of this white paper is to examine the benefits and limitations of shore based marine radar for waterside security applications. This technology has been deployed for over 60 years in Vessel Traffic Service (VTS) systems, and is more recently being used in security applications. When radar is combined with tracking and rule processing software in a security system, wide area surveillance can be automated, allowing facilities to effectively enhance waterside surveillance and meet the intent of security regulations without increasing their security staff size. Radar Based VTS Systems Marine radar has been used on land for VTS systems since 1948, when the first system was deployed in Douglas, Isle of Man. Several months later, trials were conducted in Rotterdam, and once the technology was proven, a number of European ports deployed radar based VTS systems in the 1950s. Today there are over 500 VTS systems in service globally. The VTS industry has given us much insight into the strengths and limitations of shore based radar as applied to security systems. Figure 1: VTS test site in Rotterdam, 1948. Source: Photo collection D. Zwijnenburg, VTS Manual 2002 www.tsallc.net 1
Radar Based Security Systems In the past 10 years as the interest in port security has increased, shore based marine radar is also being deployed in wide area surveillance systems. Some of the larger companies providing these systems are L-3, Lockheed Martin, and Honeywell. Smaller companies like Observation Technologies and SSR are also deploying some systems, and VTS companies such as Tideland and Kongsberg are repositioning their VTS systems toward security solutions. The number of shore based radar security systems is much smaller, but substantial growth is seen in the next 5 to 10 years as an efficient alternative to waterside surveillance using solely video analytic solutions. A common question by ports is can we use our VTS radars for security? The answer is that it s not practical to do so, because the missions of security and VTS are different, and so the settings of the radar need to be different as well. Even if a compromise configuration were implemented to support both, both VTS operators and security personnel need to have the option of changing the configuration to respond to a situation, which may impact the other mission. Benefits of Radar Based Waterside Security Systems The primary benefits of a radar based waterside security system are 1) automation of wide area surveillance, 2) improved surveillance effectiveness over other technology such as CCTV, and 3) cost savings in personnel, equipment, and maintenance. Each of these benefits will be addressed below, as will system requirements needed to achieve them. Automation of Wide Area Surveillance Automating surveillance is a force multiplier for a security team because the system works 24 hours a day, 7 days per week to monitor selected areas and only alerts an operator when there is a potential threat. The radar itself will not provide this benefit, however. The radar data must be processed to filter out noise and to determine when there is a potential threat. Once a threat has been identified, an operator needs to be alerted, and the information needed to assess the threat needs to be presented to the operator. Typical components of such a system are shown in figure 2 below. AIS Display System Radar Sensor Tracker Rule Processor Alarm Response CCTV Response Figure 2: Data flow of an automated surveillance system www.tsallc.net 2
Radar Sensor The radar sensor is marine radar that is modified for waterside surveillance. The power of the radar can range from 4KW to 25 KW, and the waveguide length of the antenna can range from 6 to 9 feet or more. The radar must be equipped with hardware to convert radar video data to raw image data that can be transmitted to tracker software over a network. Tracker The tracker software is a component of the system that receives raw radar data and extracts the radar returns that are most likely to be objects of interest. Returns that are filtered out may be from land, structures, waves, and birds. Returns that are typically of interest are from boats. Important features that tracker software must have in order to provide this processing are the ability to: Define an area of interest Exclude reflections from land Define the number of returns required to start or stop a track Allow the operator to tune the radar Send track data to a rule processor Rule Processor The rule processor receives track data from the tracker, and decides which of the tracks may be a threat. This decision process is based on user defined rules, which allow the system to differentiate between normal activity and potential threats. One rule may be that if a vessel has an AIS transponder, that it is not a threat. The rule processor must also be able to send data to a display system to generate an alarm and to initiate a camera response other wise known as slue to queue. A summary of rule processor capabilities is: Ingest track data from one or more radar sensors Ingest data from an AIS receiver Prioritize track data based on user defined rules Send display data to the display system Create an alarm response when a threat is determined Prioritize tracks and assign cameras to the threats Send camera pointing commands to the CCTV system Display System The display system provides situational awareness to the operator, usually by showing potential threats and CCTV images over a reference image or map. In this way the operator can see the location and movement of threats, and can quickly assess them with a live CCTV image. www.tsallc.net 3
Alarm Response The alarm response component receives alarms from the rule processor and displays the alarm to the operator. The alarms are stored and sorted, and in some systems can initiate an automated response. CCTV Response The CCTV response receives camera control commands from the rule processor, and uses them to automatically control one or more cameras to point to the threats. Effectiveness of Radar vs. CCTV CCTV surveillance has an inherent trade off between range and field of view. A typical security CCTV camera may have a 45 field of view at wide angle, but a short detection range of 250 meters or less. The camera can be zoomed to support a range of say 1000 meters, but at that range the field of view may be only 2. A radar sensor does not have this trade off. Radars can detect small boats at long ranges, up to 6 or even 12 Nautical Miles (NM), and have a 360 field of view. CCTV is often combined with Video Analytics (VA) to automate surveillance. While VA can generate good results in controlled environments, it is not as effective in an outdoor wide area because of the changing lighting and sea state conditions. While radar is more effective as a sensor because of these two trade offs, a pan tilt camera is still required as an assessment tool. The radar and processors can detect and initially identify what may be a threat, but ultimately an operator must make a decision based on a CCTV image to verify friend or foe. Radar can replace the detection cameras in a CCTV system, but not the assessment cameras. Cost Savings of Radar Systems An automated radar surveillance system can provide substantial savings in the cost of other technologies in equipment, infrastructure, maintenance, and personnel. Equipment costs are lower because one sensor can cover such a large area. For instance, a single radar set to a modest range of 1 NM can cover a circle area of 3.14 square NM, or over 100 million square feet. A 1000 meter camera with a 2 field of view, on the other hand, will only cover about 18 thousand square feet. Using this example, the radar can cover the area that over 5000 cameras can. As a further example of the effectiveness of the coverage of radar in the example below, 24 cameras were required for waterside surveillance coverage using video analytics, but only one radar and one long range camera was required to effectively cover the port area depicted in Figure 3. www.tsallc.net 4
Figure 3: Example of video analytic CCTV coverage vs. radar surveillance 24 Cameras for detection 1 radar for detection, 6 cameras for verification Partial coverage Full coverage Extensive set up, tuning Less set up and maintenance Limited operations in weather and light 24/7 operations Infrastructure costs are lower because fewer sensors are required, which mean power and communications network components have to be brought to fewer locations. Maintenance is less with a radar system because if there are fewer sensors, less communications infrastructure to maintain, and typically personnel costs are lower because an automated radar system will allow personnel to do tasks other than monitoring the security system until there is an alarm. Limitations of Radar Based Waterside Security Systems Limitations in a radar surveillance system are the same limitations in many electronic surveillance systems. They are probability of detection, nuisance and false alarms, resolution and discrimination, and dropped tracks. Probability of Detection (Pd) is the probability that an object of a predefined size (say a small boat of 8M or larger) will be detected in an area of interest. Pd can be measured at several points in the data flow shown in Figure 2. There is a probability that an object will be in the raw data sent from the radar to the tracker. This is controlled by the sensitivity settings of the radar. There is a probability that an object will be tracked by the tracker and sent to the rule processor. This is controlled by the tracker settings and there is a probability that an object will cause the rule processor to trigger an alarm. This is controlled by the rules that have been defined by the user. For the purposes of this paper, System Pd is measured at the display system, meaning that the radar has to detect the object, the tracker has to track it, and the rule processor has to send it to the display system. www.tsallc.net 5
System Pd is controlled by the system sensitivity settings, which are a combination of the radar settings and the tracker settings. The sensitivity desired varies from site to site, depending on the priorities of the user. If reducing nuisance alarms is a higher priority than detection probability, then the sensitivity is set low. If detection is the higher priority, then sensitivity is set high. Nuisance Alarms and False Alarms are caused by radar reflections that do not indicate objects of interest, but were not filtered out by the tracker or the rule processor. The Nuisance Alarm Rate (NAR) can increase dramatically as weather conditions change, and require the user or the software to adjust the sensitivity lower. The False Alarm Rate (FAR) is the rate not at which the system indicates objects of non-interest, but objects that do not exist at all. An example of this is a multi path radar reflection that shows an object where there is none. Radar resolution is the smallest area that a radar can look at, driven by range resolution and beam width. Radar resolution varies by range and determines how well the radar sensor can discriminate between objects that are close together. For instance, a short range radar may have a resolution of 12 meters, meaning if a boat gets within 12 meters of land or another boat, their reflections merge. This opens the possibility that a small boat may go undetected while traveling close to shore, because the shore is being ignored by the system. Objects that are being tracked may not always reflect the radar signal with the same intensity, and it is up to the tracker to decide when a track should be dropped. One parameter often used is how many sweeps with no return. As this parameter is increased, the probability of dropping a track is reduced. The negative side effect is that when a target really leaves the area, the display will indicate it s still there until the number of sweeps passes. A compromise is generally used that will allow some dropped tracks, but also dispose of stale tracks in a timely manner. The following table identifies some common limitations of a marine radar surveillance system, and the symptoms and impacts that can be expected. Limitation Symptom Description Operator action Weather Nuisance Rain and waves can cause false To avoid false positives, turn Alarm positives. down sensitivity during bad Bounce False Alarm Large vessels traveling through a channel can produce a false positive behind the vessel. This is caused by a double-bounce effect where the radar return bounces off the shore, back to the boat, and then to the antenna. Wakes Nuisance Alarm Boat wakes can produce tracks. www.tsallc.net 6 weather. Can cause a false indication on the screen behind the vessel. It should not interfere with camera pointing since the shadow is at the same bearing as the vessel. The interrogation camera should be used to verify that the false alarm is just that, a false alarm. This can cause the interrogation camera to track behind a boat rather than the boat itself. The operator can see that it s a wake and should manually steer to the boat for verification of friend or
Turning False Alarm A fast turning boat can cause a new track to be formed if it turns fast enough to leave the area predicted by the tracker. Dropouts Resolution Hiding Momentary loss of track Merged target Missed target Dropouts are common in all radar systems and can increase if the system sensitivity is set too low or if the track parameters or incorrect. Two boats very close to one another produce only one track. A smaller boat behind (parasitic vessel) a large one produces only one track. foe. This is not a common occurrence, but if it happens the interrogation camera is still pointed toward the boat and can be manually adjusted if desired. Alarms have been generated and the camera is still pointed to the object. Visual verification of the vessel s presence has been verified by the operator on the first alarm and can continue to be monitored if the vessel is a threat If the boats are close enough to be one track, then they are typically close enough to be captured in the same image. This is issue is mitigated by the operator looking at the CCTV image. It is possible for a smaller vessel to go undetected in this scenario. The mitigation is to use CCTV cameras. www.tsallc.net 7
References BRIGGS, J. N.: Target Detection by Marine Radar (The Institution of Electrical Engineers, 2004) IALA AISM: IALA Vessel Traffic Services Manual (VTS Manual, 2002) SKOLNIK, M.I. (Ed.): Radar Handbook (McGraw-Hill, New York, 1970) ROHAN, P.: Surveillance Radar Performance Prediction (Peter Perigrinus for the IEE, 1983) www.tsallc.net 8