Intelligent Containers Network Сoncept

Sergej Jakovlev

, Audrius Senulis

, Mindaugas Kurmis

, Darius Drungilas

and Zydrunas Lukosius

Informatics and Statistics Department, Klaipeda University, Bijunu str. 17, LT-91225, Klaipeda, Lithuania

Engineering Department, Klaipeda University, Bijunu str. 17, LT-91225, Klaipeda, Lithuania

Keywords: Wireless Sensors Network, Security, Intelligent Containers, Mobile Security, Communication.

Abstract: In this paper, a novel approach is presented to increase the security of shipping containers transportation and

storage in container yards. This approach includes wireless sensors networks with programmable modules

to increase the effectiveness of the decision support functionality for operators’ onsite. This approach is

closely related to the Container Security Initiative and is intended to deepen knowledge in the intelligent

transportation research area. This paper examines an urgent challenge - secure of cargo transportation in

containers, i.e., how quickly it is possible to detect dangerous goods in shipping containers without

changing their tightness and hence rationally implements international security regulations all around the

world. This paper contributes to the development of new approaches of shipping containers handling and

monitoring in terms of smart cities and smart ports (for the development of the Smart Port initiative) for

ports that have higher levels of security violations. This contribution is addressed as an informative measure

to the general public working in the Information and Communications Technologies (ICT) research area.

1 INTRODUCTION

To combat illicit trafﬁcking in maritime container

transport, a good level of detection is essential, and

should be approached with advanced data-driven or

process-driven technologies. Although the process-

driven technologies are done now with a large range

of surveillance and active interrogation techniques,

active sensors that register the threats during the

transportation route and onsite might be an

interesting supplement to the battle the rising threats.

Data-driven characteristics will allow instantaneous

recombination of all possible scenarios with a high

certainty of risk detection under normal working

conditions.

The analysis of scientific literature studying the

intermodal terminal activity revealed that there are

many models helping to improve the terminal’s

operational activity, however there are no models

helping to determine which technology would be the

most rational in the terminal for security (Chang et

al., 2014). In the research of Alexandridis et al.,

2017, they analysed the international shipping

industry in order to improve the efficacy of risk

diversification for shipping market practitioners,

further security problems were addressed by

Scholliers et al., 2016. Authors discussed the

technological possibilities to improve the integrity of

containers in port related supply chains. They

suggested that the most plausible solutions are

adding monitoring equipment, such as e-seals and

tracking devices, monitoring the environment using

cameras, improved gate processes and generating

useful control information in the general security

monitoring infrastructure, also discussed by McLay

and Dreiding, 2012.

In this paper a discussion is made allowing the

reader to generally understand the variety of

technological solutions currently applied and to

understand the importance of their integration in a

common technological platform.

Regulations and standards proposed in the

Container Security Initiative (Bullock et al., 2018)

declare that the future of containerization depends

heavily on the level of adoption of new technologies

to increase cargo and processes security on all level

and during the whole trip. CSI declared the

development of an “Intelligent container” concept.

This is how the “intelligence” in brought to the

everyday containerized section of the global

transport chain and the following CSI core elements

are achieved: establish security criteria to identify

high-risk containers based on advance information,

pre-screen containers at the earliest possible point,

568

Jakovlev, S., Senulis, A., Kurmis, M., Drungilas, D. and Lukosius, Z.

Intelligent Containers Network Concept.

DOI: 10.5220/0006801305680574

In Proceedings of the 4th International Conference on Vehicle Technology and Intelligent Transport Systems (VEHITS 2018), pages 568-574

ISBN: 978-989-758-293-6

use ICT to quickly pre-screen high-risk containers

and develop secure and “intelligent” containers.

Application of most modern mobile technologies

plays an important role in maximizing the

performance, reducing the costs and risks of

intermodal containers transportation and raises the

efficiency of other transportation services in the

supply chain. WSN is a technology that can be very

useful when it is used to acquire and dispatch

collected data in wide areas. Usually, these networks

consist of different types of nodes which are

carrying different types of sensors along with

computational devices. WSN can be visualised using

active RFID system components currently used by

several countries for port transport operations, where

each network node is called a tag. These nodes

transmit data through the network to some specific

destination or the collecting tag (initiation node).

There is a wide range of literature concerning

WSNs (Truong, 2015; Anurag & Christian, 2015).

The large amount of publications revolves around

different issues: fault tolerance (Sausen, 2010),

scalability (Hoblos, 2010), sensor placement

(Bulusu, 2001), caching and power consumption

(Dimokas, 2011), data aggregation (Polastre, 2004)

and data gathering (Krishnamachari, 2002). In some

particular cases, a WSN can be also seen as a

collection of different sensor nodes with intermittent

connectivity, asymmetric bandwidths, long and

variable latency and ambiguous mobility patterns.

There are many studies that approach the problem of

high connectivity in wireless networks.

2 ANALYSIS OF AN

INTELLIGENT CONTAINERS’

NETWORK

From the middle-ages scientists tried to mimic the

functionality of the surrounding nature by inventing

new materials and machines. Computer systems are

now control-ling various crucial aspects of our lives.

Despite this scientific leap forward, many areas of

engineering are still missing this innovative touch,

mostly due to the lack of understanding of the

benefits which can be derived from their full

integration to solve the most obvious security

problems. To adapt the intelligent container

approach to the working conditions a new method is

proposed to connect the intelligent containers to a

network with the capability to perform

computational tasks in different parts of the network

(in nodes), much like in a living brain. A container

yard can now be presented as a form of virtual brain

for a certain computational activity. In a living brain,

connection between neurons is made using nerve

tissue.

Such connection can be done using simple

cables. But this would pose serious problems to

engineers and operators’ onsite. A plausible solution

is to use wireless communication technologies to

connect all the computational neurons in the

network. Such technology is called WSN. WSN in

common applications use Ad-Hoc routing protocols.

Routing is meant to establish a proper

connection among the nodes in the network. Such

connections are fast and agile. In dynamic

environments other routing protocols may also be

applied. Each computational neuron can be

presented as an individual container with the

capability to compute certain amount of incoming

sensor information and transfer it through the

container network using wireless communication

principles. But it is a tricky problem, because where

one communication frequency is allowable in one

country, others are not.

Neurons die and the brain is evolved by

introducing new neurons and interconnections. New

containers are introduced to the stack and to the

network on a constant basis. Wireless sensor

networks or WSNs are networks of autonomous

sensors aimed at monitoring physical or

environmental conditions and pass their data through

the network to some locations or data sinks. Every

node has a radio transmitter and a limited source of

energy. Energy consumption is not essential in this

research as larger batteries may be equipped in these

conditions and may be used for years to come

without any recharge. It is possible to use a

combination of several routing protocols or a unique

protocol divided among several inner networks for

each individual case. However, there exists

paradigm that does not allow the full and effective

integration of this technology. That is the direct

communication.

At some point, using a direct communication

protocol, each sensor sends its data directly to the

base station without additional data improvement at

each node. If the base station is far away from the

nodes, direct communication requires a large amount

of resources from each node and the final result may

contain information errors. When communication is

done in a container yard, then due to the working

environment constraints, this procedure becomes

virtually impossible. Signal reflections will take

place when using ISO certified standards for

communication within the port environment.

Intelligent Containers Network Concept

569

Additional information errors in the messages will

result in additional message replies and resends.

This will take time and no guarantees are given

whether the final result will be positive.

New regulations will have to take place. One of

the solutions is to use a globally certified high

frequency ZigBee standard. Although this high

frequency standard will make transmission of

information to the initiation node (sink) complicated,

it is designed to be used in industrial environments

by using the minimum-transmission-energy routing

protocol. In this protocol, data is sent to the base

station through intermediate nodes. Nodes act as

routers for other nodes and transmit their data by

adding their own data packets. Additionally, this

data can be modified at each node separately and

resent. In other words, it is possible to correct the

data at each container node if this functionality is

programmed. Each node then can receive data from

several nodes around it at the closest distances and

make assumptions about the security of its contents

and the surrounding area. Specific hardware and

software tools should be used to reach this goal.

Additional middleware will programmable agent

logic must be introduced. Intelligent agents will act

as decision support actors for operator’s onsite and

the managers responsible for the transport chain

activities indicating possible detected threats. Each

node’s computational power can be used to assess

the problem using specific algorithms (agent logic).

These algorithms may be programmed as smart

software agents in the network and etc. The problem

still remains how to select each individual node in

the network of containers. Which sensors data is

essential and valuable and how many intelligent

containers with sensors are required to fully cover

the container yard, these are the main problems

faced in this research. The placement of the sensors

in the container can be optimized. Unnecessary

nodes can be put to sleep. Sleep functionality is

optional and can prove to be a useful toll in saving

energy. Sleeping procedure is installed in many

industrial WSN systems.

2.1 Data Communication Method

The foremost idea of the integration of the WSN and

middleware agent approach is to ensure the

collective security of the entire network in terms of

data security within the information flow for the

transport chain. To achieve this goal, an improved

method of Jakovlev et al., 2012 is proposed that

incorporates both network and hardware units, along

with the computation power of each node in the

network. It is done to increase the collective security

of the entire network of containers. In a container

yard each intelligent container performs its own

evaluation of the situation. Accuracy of these

measurements is questionable and requires further

analysis. Each container manages an area in a 3-D

space where each intelligent container can ask the

neighbouring container for assistance in data

confirmation and sensor work time efficiency

optimization. As mentioned previously, nodes can

use large power supplies and storage units inside the

containers. To increase data and autonomous process

visibility in transport chain, integrated database

should be used to store and reallocate useful

information. To achieve the security goal, a specific

messaging technique is required. When node k

receives the highly deviated sensor data it computes

the problem area Ap and initiates the request and

reply procedure. The nearest node k+1 is defined by

its coordinates x, y, z in the problem area Ap (see

figure 1). When node k sends the request message

through the network to the local data storages in

nodes and the integrated network database, the

nearest node k+1 receives that message and replies

to node k by sending the reply message m

. The

local data storages are used to store the request

messages and main network messages m

. with the

appropriate information regarding the sensor data.

Integrated network database is used to store the final

message m

. The initial request message m

, sent

from node k, and reply message m

, sent from node

k+1 at time tm, are described as (1) and (2):

{}()

,,,,,: CAptkScRtm

∈

(1)

{}()

CzyxtkScRym

,,,,,1,: +∈

(2)

Here: Ap is the problem area defined by the node

k, Rt is the set of messages sent from node k, Ry is

the set of messages sent from tag k+1, Sc is the

security mechanism and C is the message content.

When the node k+1 is found, the local data update

process is initiated. It is defined by the appropriate

network infrastructure and is used by the set of

network nodes. The data update message, sent from

node k to node k+1, is described as (3):

{}

























∈ C

zyxtTrM

QQk

ScLm

OUTkINk

,,,,,

,,,

(3)

Here: L is the set of messages sent from tag k, k

is the initial node identification number, M is node k

parameter deviation, Tr is the time of data

evaluation. The main network message m

is formed

VEHITS 2018 - 4th International Conference on Vehicle Technology and Intelligent Transport Systems

570

in node k+1 and transferred via the pre-defined

route. It is stored in the real-time communication

integrated network database. The message is

described as (4):

{}

,,,,,

,,,1

,,,,

,,,,,





































∈

zyxtTrM

QQk

Apzyxt

TrMQQk

OUTkINk

(4)

Here: S is the set of main messages m

sent from

the problem area Ap. Each message content C must

include all the general information regarding the

sensor parameters monitored by each individual

node on demand by its neighbour. Each message

content C must include all the general information

regarding the sensor parameters monitored by each

individual node on demand by its neighbour.

Figure 1: Intelligent container connectivity scheme.

This message must contain the accuracy of its

estimation. Accuracy areas within a problem area of

the container are spherical in every direction from

the initial container k (see figure 2). Each message

content C can be different, because various sensors

and their manufacturers are used by different nodes.

Their standardization requires them all to have

standard output possibilities. This is still a problem

that may not be solved at the near future. Each

message can computed specifically for the network.

Therefore, an additional security mechanism can be

placed not in the message heading but in the

message itself. The key for the message encoding is

distributed by the operating company in the network

of containers onsite using a simple network system

coding principle with a single container during its

unloading procedure.

Figure 2: Accuracy area description for the container

network.

This is done autonomously nowadays using e-

Seal technology. Distance between nodes can be

estimated in the WSN using the received signal

strength indicator parameters RSSI (Wanga, 2009).

2.2 Description of the Accuracy Area

A special accuracy area parameter is used. Accuracy

areas are designated spaces with dynamic

environment where connection is possible by

calculating the minimum SNR. Firstly, the accuracy

area is calculated and nodes are discovered. Sensors

data and parameters are sent for inspection to the

initiation node. The networking time parameter tn,

spent for both routing, accuracy area initiation and

sensors parameters data retrieval operations, is

provided. The computation time tc spent by the

initiation container k includes: time spent for inner

data file transfer, time spent for comparison to

analyzed data from inner sensors and time spent to

other sensors from outer containers. Further

communication between more than 100 nodes can

prove to be a difficult problem for an error increase

in geometrical progression. Therefore, it can be

assumed that the communication problem or tc can

be minimized by decreasing the number of

containers in the accuracy area, decreasing the range

of the accuracy areas or simply avoiding using too

many containers for communication.

As an example, this intelligent containers

network can be used to detect radioactive isotopes

(dirty bombs) in all stored containers. Additional

sensors data fusion technique like DAI-DAO can be

Intelligent Containers Network Concept

571

programmed within the agent logic to partially deal

with the communication problem by eliminating

additional noise in the sensors readings and

optimizing the overall detection time (Jakovlev et

al., 2017).

By utilizing a sensors data fusion technique for

radiation monitoring on short distances, detection

time or threshold can be shorten in comparison to

using all individual containers data separately by

individual nodes. Data can be acquired and tested for

accuracy along with the decrease in the estimation

time, when only 1 additional communication is done

for each separate accuracy area. This can prove to be

the best solution. Each container can perform its

own evaluation of the surrounding area at any given

time and thus shorten the overall inspection time for

the whole network. Ladder approach can be applied

to deal with this problem as well. Figure 2

demonstrates how communication with only one

container (k to k+1) in the highest accuracy area can

lead to estimation of the whole container yard.

Depending on the statistical background of the

evaluation, each statistical area or alert area can be

distinguished simply by estimating the distances

between containers or any other well-known and

computable onsite parameter. In this case, distance

value can be used as a unique parameter, because it

incorporates both signal strength indicator values

and indicates the nearest neighbors in the network.

In addition, background noise estimations and the

deployed sensors density values must be known as

well. Each message must contain this crucial

information. The problem still exists - which

parameters are vital for the working stability and

security. This problem can be formulated as a

rucksack problem. Each new accuracy area in 3-D

space can be formulated as (5):

)).,((_

SLH

mmdfareaAccurasy =

(5)

Messages can also include other estimations:

accuracy of background noise estimation for each

individual sensor and initial threshold time for the

estimation of true detection time. Risk levels of the

container can be estimated by different decision

support systems, like the Automated Targeting

System (ATS). Evaluation of the risk level can be

used to estimate the accuracy area as well. Its initial

value can be transferred through the network to the

initiation node. Then the accuracy area can be

evaluated taking into consideration the importance

of the specific container. Further risk level

assumptions are made within the initiation node. The

final decision is made by the node to increase the

accuracy area to perform additional evaluations of

the risky container. These parameters may vary

differently for each individual container. If no pre-

determined risk is assessed by the ATS, then the

priority of each node in the stack is the same. Each

individual accuracy area H can be separated in-

between other areas by a default value of 5%.

(i.e. 0…100% with a 1…5% step). This principle is

applied when a certain degree of accuracy is

computed by the smart agent using the data defect

levels.

The container accuracy area evaluation method

includes two interconnected parts: the container

itself and its built-in sensors. The reliability of each

WSN node data is examined by the smart agent and

the reliability of the information is examined. The

decision support functionality work as an expert

system within a node that provides decision about

the truthfulness of the monitored container status.

They include the following suggestions: each

container can be in normal or defected state and

network sensors can also be in normal state and

provide correct information or in a defected state

thus providing false information. Data is considered

to be reliable when they describe the actual state of

container.

The defect levels for WSN node k and the

nearest WSN node k+1 are introduced: Q

INk

is the

defect level of the incoming data in node k (any

node in the network),

()

1,0∈

INk

; Q

OUTk

is the

defect level of data after the evaluation in node k,

()

1,0∈

OUTk

;

1+INk

is the defect level of the

incoming data from node k which is transferred to

the neighbouring node for potential evaluation,

()

1,0

∈

+INk

and

1+OUTk

is the modified defect

level of data after the secondary evaluation in node

k+1 (nearest node in the network),

()

1,0

∈

+OUTk

In all cases, the defect level Q

OUTk

is defined as

INk+1

for the nearest node in the network during the

data update process. This is done in order to check if

the acquired data is true or false to be used further in

the estimation of the trustworthiness of the

containers and accuracy area. The full evaluation of

the accuracy area can be rewritten as (6):

.100~_

⋅

+OUTkH

QareaAccurasy

(6)

In this method, each intelligent container can

provide its own observation according to the

requirements that are coded within the middleware

agent. Each container node can compute its crucial

parameters and then ask for other nodes to do the

same. A step-by-step solution is presented (figure 3).

VEHITS 2018 - 4th International Conference on Vehicle Technology and Intelligent Transport Systems

572

Figure 3: Overview scheme of the step-by-step solution.

One may notice that it is not an optimal solution.

In this case t2 and t4 describe the networking time

for the communication and messaging, t1, t3 and t3

are the computational times. Then the sum of the

computational and the networking time values can

be presented as

54321

tttttt

Step

++++=

. In this

case, information acquisition for node k is done in a

step-by-step manner. This step-by-step method can

be changed according to the network activity. Each

network node can acquire all the needed information

in advance, process and store it locally. The

proposed method can be used to avoid data loss in

the network, when many nodes are talking to each

other simultaneously. This could crash the entire

network if a suitable synchronization protocol is not

adapted. An obvious solution would be to integrate

both methods into a single procedure. Monitoring of

the environment can be done simultaneously to

decrease the total detection time in the entire

network and messages routing in the network can be

done step-by-step at a local accuracy area pre-

determined using the same principle for each

neighbour. In figure 4, a different approach is

presented.

Figure 4: Overview scheme of the simultaneous network

nodes activity.

This simultaneous network nodes activity

method shows that in time

321

tttT

Sim

++=

information can be gathered more quickly, taking

into account all the other actions related to time

taken for sensor data manipulation by the agent,

network routing and data transmission.

Simultaneous messages retrieval can also trash the

network and make it unstable.

3 CONCLUSIONS

In this proposed method, each node performs a local

decision support based on the prediction of the

background noise, estimation of the accuracy of the

estimation and its surrounding area. The estimation of

the required data sample size for the initial communi-

cation is a serious mathematical and computational

problem, because each individual scenario requires a

different statistical analysis approach for computing

data reliability. The integration is possible only when

there all necessary standardization tasks are finished

and the system is widely used throughout the

transport chain. This innovation must be taken into

consideration not only by a single port authority, but

by the whole global transport chain.

Therefore, any intelligent container knows the

exact info it needs to know at the most appropriate

moment and predict its neighbour’s possible

deviations in the monitored spectrum. This

functionality is already implemented in some E-Seal

systems. As briefly mentioned previously,

application of intelligent systems plays an essential

role in achieving the optimality goal of security in

many countries of the world. These networking

technologies can be applied in both in container

yards, trucks, trains and ships to connect each

individual container in a common network.

Future work includes research on the impact of

delays, errors and other uncertainties on the

communications protocol, its application in

laboratory environment and in practice using

research grant described below.

ACKNOWLEDGEMENTS

Authors would like to thank EU funded Research

project “Ateities autonominis žaliasis uostas: naujo

konteinerių krovos metodo ir sistemos prototipo

sukūrimas“ (Nr. 01.2.2-LMT-K-718-01-0081) for

the support while writing and publishing the

manuscript.

Intelligent Containers Network Concept

573

REFERENCES

Chang, C.H., Xu, J., Song, D.P. 2014. An analysis of

safety and security risks in container shipping

operations: A case study of Taiwan, Safety Science.

Vol. 63, p. 168-178.

Alexandridis, G., Sahoo, S., Song, D.W., Visvikis, I. 2017.

Shipping risk management practice revisited: A new

portfolio approach, Transportation Research Part A:

Policy and Practice, In press, corrected proof.

https://doi.org/10.1016/j.tra.2017.11.014.

Scholliers, J., Permala, A., Toivonen, S., Salmela, H.

2016. Improving the Security of Containers in Port

Related Supply Chains, Transportation Research

Procedia. Vol. 14, p. 1374-1383.

McLay, L. A., Dreiding, R. 2012. Multilevel, threshold-

based policies for cargo container security screening

systems, European Journal of Operational Research.

Vol. 220(2), p. 522-529.

Jane A. Bullock, George D. Haddow, Damon P. Coppola.

2018. 7: Transportation Safety and Security. Book

chapter. Homeland Security (Second Edition),

ELSEVIER. p. 169-188.

Truong, T.T., Brown, K.N., Sreenan, C.J. 2015. Multi-

objective hierarchical algorithms for restoring

Wireless Sensor Network connectivity in known

environments, Ad Hoc Networks. Vol. 33, p. 190–208.

Anurag, S., Christian, W. O. 2009. Performance

comparison of particle swarm optimization with

traditional clustering algorithms used in self-

organizing map. International Journal of

Computational Intelligence. Vol. 5(1), p. 32–41.

Sausen, P. S., Spohn, M. A., Perkusich, A. 2010.

Broadcast routing in wireless sensor networks with

dynamic power management and multi-coverage

backbones. J. Inform. Sci. Vol. 180(5), p. 653–663.

Hoblos, G., Staroswiecki, M., Aitouche, A. 2000. Optimal

design of fault tolerant sensor networks. IEEE

International Conference on Control Applications,

pp. 467–472.

Bulusu, N., Estrin, D., Girod, L., Heidemann, J. 2001.

Scalable coordination for wireless sensor networks:

self-configuring. International Symposiumon

Communication Theory and Applications, p. 1-6,

Ambleside, UK.

Dimokas, N., Katsaros, D., Tassiulas, L., Manolopoulos,

Y. 2011. High performance, low complexity

cooperative caching for wireless sensor networks. J.

Wirel. Network. Vol. 17(3), p. 717–737.

Polastre, J., Hill, J., Culler, D. 2004. Versatile low power

media access for wireless sensor networks.

Proceedings of the 2nd International Conference on

Embedded Networked Sensor Systems, p. 95–107.

Krishnamachari, L., Estrin, D., Wicker, S. 2002. The

impact of data aggregation in wireless sensor

networks. 22nd International Conference on

Distributed Computing Systems Workshops. p. 575 –

578.

Wanga, X., Bischoffa, O., Laura, R., Paula, S. 2009.

Localization in Wireless Ad-hoc Sensor Networks

using Multilateration with RSSI for Logistic

Applications.

In: Proceedings of the Eurosensors

XXIII conference. Procedia Chemistry. Vol. 1, p. 461–

464.

Jakovlev, S., Andziulis, A., Bulbenkienė, V., Didžiokas,

R., Bogdevičius, M., Plėštys, R., Zakarevičius, R.

2012. Cargo Container Monitoring Data Reliability

Evaluation in WSN Nodes. Electronics & Electrical

Engineering, Vol. 3(119), p. 91-94.

Jakovlev S., Kurmis M., Drungilas D., Lukosius Z.,

Voznak M. 2018. Multi-sensor Data Fusion Technique

to Detect Radiation Emission in Wireless Sensor

Networks. AETA 2017 - Recent Advances in

Electrical Engineering and Related Sciences: Theory

and Application, Lecture Notes in Electrical

Engineering, Vol. 465, p. 135-144.

VEHITS 2018 - 4th International Conference on Vehicle Technology and Intelligent Transport Systems

574