**2.1 The iCrowd simulator**

estimate risk from both Observable and Hidden risk indicators across all four security tiers and heavily tested, both in vivo and in vitro through simulation, in carefully designed pilots across all three BCP modalities: air, land and sea.

projects and pilot use-cases.

*TRESSPASS comprehensive risk-based security framework.*

**Figure 11.**

*Deep Learning Applications*

**Figure 12.**

**124**

*Photo-realistic, agent-based simulation using iCrowd.*

**Use of simulation in designing, testing, and assessing risk-based security** Paramount to the design and testing alternative designs of risk-based security concepts, technologies and protocols, in order to achieve the increase in effectiveness of BCPs with the parallel reduction of delays, is the use of simulation. iCrowd is an agent-based simulator that can be used to implement and test different riskbased concepts and technologies in a flexible and realistic simulation environment [6]. **Figures 12** and **13** show a photo-realistic virtual reconstruction of an airport used extensively in simulating security scenarios and policies for a variety of

The iCrowd Simulator is an agent-based simulation platform capable of handling small-scale to large-scale crowds and calculating the change of the status of each participating component depending on dynamic interactions with other entities or the environment during simulation time [7, 8]. It can be utilized in any bounded area, i.e. building interiors and exteriors, stadiums, or any exterior area e.g. public places like squares, open-air festival etc. Currently, it is being used to simulate crowd movement and crowd interactions in general, with the graphical display being optional. As *an agent-based simulation platform, different parameters for each agent can be considered, such as physical, emotional,* vital characteristics regarding the crowd that will be observed (i.e. stress levels, health status), object/obstacle parameters and also environmental parameters that can affect the final solution of each simulated scenario performed.

#### **Figure 14.**

*(a): Aspect of third-person camera. (b): Path planning example: The green line indicates the path the selected agent (displayed are red) is following (c): Travelers enter the airport. The display of hold and hand luggage is turned on. (d): Travelers go through the check-ins. The display of hold and hand luggage is turned on.*

iCrowd offers a fully operational flow simulation for travelers and personnel inside an airport, as displayed in **Figures 12** and **13**. It enables the user to define simulation scenarios, it is implementing a sophisticated crowd engine with collision avoidance6 with multiple, different behaviors that can co-exist inside the same simulation. It also supports distributed simulations, operating as an orchestrator. It has been integrated with the C2 Web Portal OCULUS Air to communicate data, such as displaying the position and the movement of simulated entities in real time, and the Fusion and Ingestion Server to update travelers'status accordingly depending on their interactions with the airport's hardware and other security technologies (i.e. Beacons, RFID scanners and RFID tags for carry-on luggage tracking), **Figure 14(a)**–**(d)**.

discussed above, we assumed that travelers can be tracked anonymously using topdown view cameras in compliance with GDPR and ethics regulations. Based on a model of what constitutes a normal traveler route (trajectory) in an airport (or similarly any other BCP), a convolutional recursive neural network was trained with "normal trajectories" generated by the iCrowd simulator. Once the RNN is trained with "normal trajectories," travelers with "suspicious behaviors" are generated among travelers with "normal behaviors" and the algorithm is tested if it could detect the "suspicious trajectories." In **Figure 14(b)**, the traveler with suspicious behavior is color-coded red. The risk assessment algorithm detects and identifies

*Risk Assessment and Automated Anomaly Detection Using a Deep Learning Architecture*

A complete technical description of the anomaly detection algorithm is given in

the references [9, 10]. Next, we summarize the results in [9, 10] in order to demonstrate the possibility of implementing a risk-based security system that monitors traveler risk continually without additional delays that can offset the

The evaluation of the risk assessment system of [9, 10] is done using the Precision-Recall (PR) diagram, the Receiver Operating Characteristic (ROC) curve,

• Precision = (# of true suspicious behaviors detected)/(# of total labeled

• Recall = (# of true suspicious behaviors detected)/(# of total suspicious

• Receiver Operating Characteristic (ROC) curve = Probability of detection

• F1-score = 2\*[(Precision \* Recall)/[Precision + Recall] which is the harmonic

**Figures 15**–**17** illustrate the PR diagram, the ROC curve and the Confusion

respectively. It should be reminded that the values of all evaluation metrics are defined within the interval [0, 1]. The closer to 1 a value lies, the better the achieved performance. **Table 1** summarizes the values of the recruited evaluation measures. The threshold score derived by the RNN architecture, by maximizing

Furthermore, the F1 score (i.e. the harmonic average between Precision and Recall, Eq. (1)), along with the Total Accuracy, Eq. (2), the ROC AUC (Area Under

*<sup>F</sup>*<sup>1</sup> � *score* <sup>¼</sup> <sup>2</sup> � Pr*ecision* � Re *call*

*Total Accuracy* ¼ ð Þ #*of Successful Assessments =*ð Þ #*of Total Cases* (2)

Pr*ecision* <sup>þ</sup> Re *call* (1)

• Confusion Matrix = Normal versus abnormal confusion matrix

• Total Accuracy = (# of Total Assessments)/(# of Total Cases)

Matrix respectively. Eqs. 6 and 7 calculate F1-score and Total Accuracy

Curve), and Average PR score, are calculated in **Table 1**,

the Confusion Matrix, the F1-score and the Total Accuracy, as defined next:

the suspicious traveler in **Figure 14(d)**.

*DOI: http://dx.doi.org/10.5772/intechopen.96209*

benefits of the risk-based approach.

suspicious behaviors)

versus false alarm probability diagram

average of Precision and Recall.

*3.1.1 Evaluation results*

behaviors)

F1-score, is 3.7.

**127**
