Monitoring a Sensor Output to Determine Intrusion Events

Abstract

A method of detecting intrusion events including at least two different event types which have different characteristics of frequency and time comprises providing a sensor responsive changes in a medium generated by a potential intrusion event with the sensor generating an output signal indicative of the changes in the medium, analyzing the signal to determine changes in amplitude so as to detect the change in amplitude of the detection signal as a function of time, and performing at least one of: (i) in the frequency domain, carrying out a frequency analysis of the signal from the sensor and dividing the frequency analysis into separate sections which are selected so as to correspond to the characteristic frequencies for each event type, or (ii) the algorithm requiring the presence or absence of a time domain step function.

Claims

1. A method of detecting intrusion events comprising: wherein the intrusion events include at least two different event types which have different characteristics of frequency and time; providing a sensor responsive to changes in a medium generated by a potential intrusion event with the sensor generating an output signal indicative of the changes in the medium; analyzing the signal in the time domain to determine changes in amplitude so as to detect the change in amplitude of the detection signal as a function of time; in the frequency domain carrying out a frequency analysis of the signal from the sensor; and dividing the frequency analysis into separate sections which are selected so as to correspond to the characteristic frequencies for each event type.

2. The method according to claim 1 wherein the frequency analysis provides a combination of events in a multi-dimensional matrix that analyzes at least one of relative amplitude of each frequency, the duration of each detected event, the repetition rate of said event, the period over which this event occurs, and the presence or absence of a time domain step function.

3. The method according to claim 1 wherein the characteristic frequencies are selected so as to allow detection and suppression of false alarms using the analyses in the time and frequency domains for the signals.

4. The method according to claim 1 wherein certain events are excluded as false alarms if they do not meet the frequency and/or time characteristics determined for the event types.

5. A method of detecting intrusion events comprising: wherein the intrusion events include at least two different event types which have different characteristics of frequency and time; providing a sensor responsive to changes in a medium generated by a potential intrusion event with the sensor generating an output signal indicative of the changes in the medium; analyzing the signal in the time domain to determine changes in amplitude so as to detect the change in amplitude of the detection signal as a function of time; wherein the analysis requires the presence or absence of a step function in the time domain.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The invention will now be described in conjunction with the accompanying drawings in which:

[0032] FIG. 1 is a graph of amplitude v time for the signal over a number of time bands;

[0033] FIG. 2 is a graph of amplitude v frequency for the bands; and

[0034] FIG. 3 is a schematic diagram of an arrangement of medium and sensor in which the method of the present invention may be applied.

[0035] In the drawings like characters of reference indicate corresponding parts in the different figures.

DETAILED DESCRIPTION

[0036] With reference to the accompanying figures, the present invention relates to detection and false alarm suppression algorithms for the signals obtained by the above techniques or signals from other sensors.

[0037] The current method for detection lies significantly in a simple monitoring of the sensor and detect threshold crossings of amplitude. This, however, offers no discrimination between different event types such as cut, climb, and wind events.

[0038] This invention is multi layered, as follows:

[0039] Layer 1 consists of two algorithmsa time domain discrimination algorithm and a frequency domain algorithm.

[0040] The time domain, at its root level, detects the change in amplitude of the detection signal as a function of time. That is, it monitors absolute change over a time slice, as illustrated in FIG. 1. FIG. 1 shows a level in decibels (dB) of the detection or output signal S over time. One key feature of this analysis is that the signal in respect to time should display a step function as shown in the Figures where the signal moves from level A to level B in a set period of time. For example, in order to be considered a step function, the level of the signal should increase by a prescribed threshold of 2 dB over a prescribed time interval of five seconds, that is when comparing the level at the beginning of the period as indicated at I and at the end thereof as indicated at II. Generally the algorithm will check whether the signal level has exceeded the threshold within the prescribed time interval. This allows the distinction to be made between the event types and the false alarms as the event type to be determined is required to meet this step function. If it does not it is either an event of type B or is neither and must therefore be a false alarm.

[0041] The frequency domain algorithm does a frequency analysis of the signal from the sensor, such as a Fast Fourier Transform. This frequency envelope is partitioned into multiple sections that correspond to the primary frequencies for each event type.

[0042] That is, prior analysis of each event type to be detected is carried out to determine time and frequency characteristics of the event.

[0043] For example, crossover points at 50 Hz and 500 Hz, as shown:

[0044] This invention utilizes a combination of events in a multi-dimensional matrix that analyzes one or more of: relative amplitude of each frequency, the duration of each detected event, the repetition rate of said event, the period over which this event occurs, and the presence or absence of a time domain step function.

[0045] As tabulated below:

TABLE-US-00001 Relative Amplitude per Presence Freq Band Scale 1-10 Event Repetition Repetition of Time F1 F2 F3 F4 FN Duration Rate Period Domain Wind 1- 1- 1- 1- 1- A Sec B Hz C Hz scale 1-10 10 10 10 10 10 Climb 1- 1- 1- 1- 1- L Sec M Hz N Hz scale 1-10 10 10 10 10 10 Cut 1- 1- 1- 1- 1- X Sec Y Hz Z Hz scale 1-10 10 10 10 10 10

[0046] For example, a person climbing a fence might step every 1.5 second, with an event lasting 500 mS, over the course of several seconds, with a heavy emphasis on the mid frequencies and presence of a time domain step function.

[0047] In another example, a person cutting the fence might show a clip every 500 mS, with an event lasting 100mS, over the course of tens of seconds, with a heavy emphasis in high frequencies and an absence of a step function.

[0048] This interaction of the data allows the system to:

[0049] 1) Send out alerts that an unknown episode is occurring on the fence as soon as a signal is received indicative of a potential event.

[0050] 2) After the appropriate time, the algorithm indicates the type of alert concerned such as cut or climb. This is carried out by the analysis herein wherein signal is analyzed for the frequency and time characteristics of the event type.

[0051] 3) The same analysis allows the analysis to exclude certain events as false alarms if they do not meet the frequency and/or time characteristics determined for the event types.

[0052] This methodology can be expanded to accommodate other alarms or variables:

[0053] The characteristics of the event types can include many or few frequency bands of potentially varying widths.

[0054] The time characteristics of each event type can include more granularity in the time domain that monitors attributes such as repetition rate and period, including a multiple step envelope function showing rise, sustain, and fall times and rates.

[0055] The arrangement herein is not limited to sensors which generate signals by optical fibers or other conducts and can use other types of sensors which generate a detectable signal in response to other detectable events such as door opening, manhole cover lift, digging a hole.

[0056] FIG. 3 schematically illustrates an example of system which can perform the method of detecting intrusion events described hereinbefore. In this example the containment barrier being monitored is a fence 1 standing upwardly from ground surface 3. A detection medium 4 for example light carried by a fibre optic cable is operatively coupled to the barrier so that so that changes in a condition of the barrier marked by a potential intrusion event, for example vibration thereof which differs from an anticipated normal stationary condition of the barrier, acts to effect changes in the detection medium 4. A sensor 5 is operatively connected to the detection medium 4 to respond to those changes to generate an output signal indicative of the changes in the medium 4. The sensor 5 also is operatively connected to a computing system 8 such that the computing system can receive the output signal for analysis. The computing system 8 generally comprises a processor 9 and a memory 10 which are operatively interconnected. The computing system 8 conducts the analysis which includes an analysis in each of the time and frequency domains. The time domain analysis is used to determine whether the output signal includes a step function which normally is indicative of a potential intrusion event. If there is no such step function in the signal then this likely corresponds to a false alarm. The frequency analysis is used to identify further characteristics of the potential intrusion event. After the time and frequency domain analyses are completed the time and frequency characteristics are compared to a predetermined matrix or data table of the same types of time and frequency characteristics of a plurality of possible intrusion events. By comparison to these values in the matrix/table it can be determined what the potential intrusion event is, or whether it is a false alarm if the characteristics derived from the analysis of the potential intrusion event do not suitably match any set of values in the matrix. The computing system 8 is further arranged for indicating to a user what type of intrusion event has been detected, including whether this is a false alarm, for example by display 12.

[0057] The scope of the claims should not be limited by the preferred embodiments set forth in the examples but should be given the broadest so interpretation consistent with the specification as a whole.