This book offers tools, modeling principles and state-of-the art simulation models for discrete-event based network simulations. Wired and wireless networks and. The need for communications network modeling and simulation. .. We use the simulation modeling tool COMNET  to depict the respective probability. Purchase Modeling and Simulation of Computer Networks and Systems - 1st Edition. eBook ISBN: . It focuses on the theories, tools, applications and uses of modeling and simulation in order to effectively.
|Language:||English, Spanish, Indonesian|
|Genre:||Business & Career|
|ePub File Size:||29.46 MB|
|PDF File Size:||15.71 MB|
|Distribution:||Free* [*Regsitration Required]|
This book focuses on tools, modeling principles and state-of-the. Buy eBook The parts dealing with modeling and models for network simulations are split. In addition, a large set of tools have evolved that support the modeling, the programming and the execution of simulation code for the evaluation of networks at. Get this from a library! Modeling and tools for network simulation. [Klaus Wehrle; Mesut Güneş; James Gross;] -- A crucial step during the design and engineering .
The increasing complexity and low-power constraints of current Wireless Sensor Networks WSN require efficient methodologies for network simulation and embedded software performance analysis of nodes. In addition, security is also a very important feature that has to be addressed in most WSNs, since they may work with sensitive data and operate in hostile unattended environments. In this paper, a methodology for security analysis of Wireless Sensor Networks is presented. After an analysis of different WSN attack types, an attacker model is proposed. This model defines three different types of attackers that can emulate most WSN attacks. In addition, this paper presents a virtual platform that is able to model the node hardware, embedded software and basic wireless channel features. This virtual simulation analyzes the embedded software behavior and node power consumption while it takes into account the network deployment and topology.
Drawing upon years of practical experience and using numerous examples and illustrative applications recognized experts in both academia and industry, discuss: Discusses important and emerging topics in computer networks and Systems including but not limited to; modeling, simulation, analysis and security of wireless and mobiles networks especially as they relate to next generation wireless networks Provides the necessary methodologies, strategies and tools needed to build computer networks and systems modeling and simulation from the bottom up Includes comprehensive review and evaluation of simulation tools and methodologies and different network performance metrics including mobility, congestion, quality of service, security and more.
List of contributors Preface Overview and goals Features of the book Organization and scanning of chapters Target audience Acknowledgements Part 1: Protocols and services in computer networks and systems Chapter 1. Wireless and mobile technologies and protocols and their performance evaluation 1 Introduction 2 Wireless and mobile technologies 3 LTE 4 LTE-advanced 5 Wireless local area network 6 Simulation of wireless networks 7 Conclusion References Chapter 2.
Rate adaptation algorithms for reliable multicast transmissions in wireless LANs 1 Introduction 2 Related work 3 The Simulation techniques for evaluating energy-efficient heuristics for backbone optical networks 1 Energy efficiency in optical backbone networks 2 Software simulator for energy-efficient optical networks 3 Object-oriented principles for design 4 Network environment 5 Exploiting traffic grooming and related heuristics 6 Adaptive methods to reduce complexity 7 A simple yet efficient heuristic: Approaches in performance evaluation Chapter 6.
Generating realistic workload for web performance studies 1 Introduction 2 Workload models and the current web 3 Web workload generators overview 4 DWEB: Computer networks performance modeling and simulation 1 Introduction 2 Performance modeling 3 Performance metrics in computer network simulation 4 Discrete-event simulation 5 Validation and verification 6 Network simulators 7 Case study: The impact of dynamic user workloads on web performance: Modeling approaches of computer networks and systems Chapter On the self-similarity of traffic generated by network traffic simulators 1 Introduction 2 Self-Similarity and the Hurst parameter 3 Self-Similarity in Network Traffic 4 Analysis of self-similarity in simulated network traffic 5 Conclusions and future work References Chapter Performances evaluation and Petri nets 1 Introduction 2 Modeling and performance evaluation 3 Petri nets 4 Conclusion References Chapter Markov chain models and applications 1 Introduction 2 Strengths of Markov models 3 Analytical modeling techniques 4 Markov modeling 5 Markov regenerative process modeling 6 Conclusions Acknowledgement References Part 4: Simulation methodologies in computer networks and systems Chapter A model-driven method for the design-time performance analysis of service-oriented software systems 1 Introduction 2 Related work 3 Background 4 Model-driven method 5 Example application 6 Conclusions References Chapter Current and future trends in open source network simulators for wireless systems 1 Network simulation issues 2 Simulation frameworks overview 3 Open source network simulators 4 OS-oriented tools 5 NS-3 frameworks References Chapter Simulating wireless and mobile systems: Simulation methods, techniques and tools of computer systems and networks 1 Introduction 2 Simulation techniques, models and tools 3 Discrete event simulation 4 GPU-based simulations 5 Multi-Agent—Based Simulation 6 Classification of some discrete event simulation techniques 7 Conclusion References Chapter An integrative approach for hybrid modeling, simulation and control of data networks based on the DEVS formalism 1 Introduction 2 Background and tools 3 Application to a problem of admission control 4 Application to a problem of congestion control 5 Conclusions References Part 5: Next generation wireless networks evaluations Chapter An Ns-3 based simulative and emulative platform 1 Introduction 2 The evolved packet system 3 Differentiated services domain 4 The Ns-3 simulator 5 The developed tool 6 Conclusions and future development References Chapter Analysis and performance evaluation of the next generation wireless networks 1 Introduction 2 The evolution of cellular wireless systems 3 Modeling and analysis of interference in the heterogeneous wireless networks 4 Simulation techniques for the next generation wireless heterogeneous networks 5 Conclusion Acknowledgement References Chapter Evolutionary algorithms for wireless network resource allocation 1 Introduction 2 Related work and algorithms 3 System model and resource allocation problem 4 Evolutionary algorithms for resource allocation 5 Select results and comparison 6 Conclusions References Chapter Modeling tools to evaluate the performance of wireless multi-hop networks 1 Introduction 2 Background 3 Performance models 4 Conclusion References Chapter Multimedia transmission over wireless networks fundamentals and key challenges 1 Introduction 2 Multimedia transmission over wireless networks 3 QoS and QoE: Software-defined wireless network SDWN: Radio resource management for heterogeneous wireless networks: Modeling and simulation for system security Chapter DoS detection in WSNs: Energy-efficient designs and modeling tools for choosing monitoring nodes 1 Introduction 2 Related work 3 Pseudo-random self-election of the cNodes 4 Modeling with markovian processes and GSPN model 5 Energy-based designation of the cNodes 6 Conclusion References Chapter Formal methods of attack modeling and detection 1 Introduction to computer system threats 2 Modeling computer system attacks 3 Analysis and detection of computer system attacks 4 Case studies 5 Conclusions References Chapter Security analysis of computer networks: Key concepts and methodologies 1 Introduction 2 Fundamental security objectives in computer networks 3 Vulnerability and malware in computer networks 4 Security threats and attacks in computer networks 5 Defense mechanisms against security attacks 6 Security analyses of computer networks 7 Emerging topics and research challenges for security analysis 8 Conclusion Acknowledgement References Index.
English Copyright: Powered by. You are connected as. Connect with: Use your name: Thank you for posting a review! Specifically, the main contributions of this paper are:. A methodology that allows the design of secure WSNs from the first steps of the development process. Moreover, using the simulator it is possible to guide the development and to verify different attack countermeasures. A WSN attack model that comprises a large number of attacks in only four categories, thus facilitating its implementation and analysis.
An attack simulator framework that is able to model, simulate and estimate the impact of attacks over different kinds of networks. The paper has been organized into seven sections. Section 3 presents the design methodology for secure WSNs. Section 4 explains the WSN simulator. Section 5 describes the WSN vulnerabilities and introduces an attack classification. This section also defines the proposed attackers.
Section 6 reports some experimental results. Finally, Section 7 states the conclusions. Nodes in Wireless Sensor Networks are usually highly energy-constrained and are often expected to operate for long periods with limited energy reserves.
For this reason, early performance estimation is an essential step in any embedded system design methodology.
Early, fast and accurate simulations can provide information to the WSN developers that enable the modification of the SW algorithms or the network architecture in order to optimize the WSN design for the best use of the limited resources. There are several simulators in the literature. In [ 19 , 20 , 21 , 22 ], authors present surveys of WSNs simulators. NS-2 provides support for simulation of Transmission Control Protocol TCP , routing, and multicast protocols over wired and wireless networks.
Avrora is a simulation tool which helps to develop sensor network simulation with clock cycle accurate execution of microcontroller programs. Avrora can only emulate two specific platforms. Peng Lei et al. In [ 35 ], UML was applied to model a specific system, consisting in measuring, pre-processing, wireless transmitting and post-processing data from sensors. Luca Berardinelli et al. Another simulator is J-Sim [ 37 ] a component-based simulation environment developed in Java.
The main limitation of this simulator is its low efficiency. The main limitation of this simulator is that it only allows the simulation of underwater networks. The approach in Shawn [ 39 ] is an open-source discrete event simulator designed to simulate large-scale networks from an abstract point of view. However, it does not support real traffic simulation and real SW code. Other simulators include Prowler [ 40 ] and JProwler [ 41 ]. They are probabilistic wireless sensor network simulators.
Prowler is written in Matlab, while JProwler is written in Java. Traditional network simulation environments do not capture the operation of endpoint nodes in detail. In most cases, the simulation does not consider the hardware and the software in each node. If this information is ignored, it is impossible to estimate the consumption of the network with precision.
Considering that increase in power consumption is one of the main problems of an attacked network, it is very important, in order to obtain valid and accurate estimations, to use a simulator that fulfills some requirements:. Use of Real Network Traffic: In order to estimate the attack effects, it is important to work with the real traffic that the network will have when it is deployed. Simulation of the Real SW Code: To calculate the behavior of the SW code against attacks, it is very important to simulate the WSN using the final SW code of each node.
HW platform support: The power consumption will vary depending on the Hardware components of the platform. The simulator must support different architectures. OS support: Nowadays, the software of an embedded system can run over an operating system. It is of great value for the simulator to support different and typical OSs. Power consumption: The estimation of power consumption is one of the main requirements to estimate the attack effects.
Encryption Security Metric: In order to detect potential vulnerabilities in wireless transmissions, it is important to measure the security of the encrypted packets. Table 1 presents some simulators considering these requirements. The main lack of these simulators is that none of them has a realistic level of simulation with different operating systems and with power consumption estimation. Power consumption estimation is one of the critical points in these systems.
Few of these simulation tools have considered power consumption. They have problems with the scalability of their hardware because they are based on a specific processor or on specific hardware. Another problem with these simulation environments is that they do not simulate the specific software deployed on them, and the network traffic is based on external functions, which is not real traffic.
As can be observed in Table 1 , the traffic generations of the simulators, in most of cases, are not based on the real traffic generated by the real applications. Normally, the traffic is generated by traffic patterns for example, in NS simulators, the traffic can be generated with different distributions, such as Pareto or exponential , Statically or Dynamically for example TOSSIM is a discrete event simulator which can be generated dynamically with external scripts or statically by default or probabilistic such as Prowler, which generates its traffic with external functions.
Our commitment is to simulate the system performance under attack situations. Without real traffic information it is very difficult, if not impossible, to perform an accurate simulation of real attacks, since their impact on the system behavior mostly depends on corner cases resulting from exploiting the weakness of the application SW. Moreover, none of these simulators provide a metric to measure the security of the encrypted transmissions. As can be observed, this simulator complies with the previous requirements.
It has SW and HW support, estimates power consumption and the traffic of the network is generated directly with the software code of each node. There is little work about attack simulation in WSNs. The network traffic is based on external functions in [ 44 , 45 ]. Another framework for the simulation of communication networks attacks is NETA [ 46 ].
The problem of this approach is that it only allows the simulation of three different attacks. Analytical models aimed at detecting and contrasting attacks are discussed in [ 47 , 48 , 49 ].
Xu et al. Kaplantzis et al. In these cases, simulation is used to validate their correctness and efficiency. Wang et al. In [ 52 ], UML Sequence Diagrams are used to describe and analyze possible attacks in a network and transport layer. It makes use of a WSN simulator that includes attack simulation. The proposed methodology is presented in Figure 1. It includes a simulator that allows the evaluation of the network behavior under different conditions different network topologies, attacks, software versions, etc.
The estimations that the simulator provides allow the detection of the most harmful attacks and the most vulnerable nodes or configurations. These evaluations of security, power and behavior also help developers to select which countermeasure is added.
The data obtained from the simulation can be used by developers to evaluate the attack effects on the network nodes. As a result, developers can design, develop or modify a custom countermeasure for each network node. Once a countermeasure is implemented, it can be tested with a new simulation. As shown in Figure 1 , in the evaluation step of the methodology, developers can explore and compare the effects of the attacks with different configurations or countermeasures.
This enables the modification of the application software or hardware of the nodes in order to improve the performance of the network. Additionally, this methodology allows the comparison of different countermeasures, thus only the most efficient are implemented in the final systems. These countermeasures can be selected from a wide range of existing techniques in the state-of-the-art or can be designed with the support of the simulator.
Different types of attacks are injected in order to evaluate the behavior and performances mainly power consumption, response time and active period of the WSN nodes. The simulation also takes into account the network deployment.
Evaluation of attacks: An attack can be very harmful for a specific node but harmless to another node. Thus, the WSN simulation will help to identify the most problematic attacks and which parts of the WSN could be most compromised. With the proposed virtual platform, it is possible to simulate a sufficiently accurate hardware and network model under attack conditions while the real embedded software is being executed in the nodes. With the simulation and performance results, it is possible to identify the most dangerous attacks for the WSN.
Select a countermeasure: The previous stage enables the detection of the most harmful attacks, thus the next step is to select and evaluate the possible countermeasures. Additionally, it is also possible to use the estimations to modify the embedded software and minimize the attack effect. For this reason, this stage includes two steps:. Design of an attack detection procedure: In order to guide this process, the evaluation of the system behavior and estimations provided by the WSN simulator are studied to find the effects that the attacks produce.
The objective is to identify a method to detect when a node is being attacked so that a solution to that attack can be implemented. For example, if the network is simulated in normal conditions without attacks a rate for the transmitted and received packets can be obtained for a particular network deployment. With the estimations that the virtual simulator provides, the developers can use the attack effect to detect the instant in which an attack takes place.
In the case of a jamming attack, the traffic rate varies compared with normal conditions. Because of this variation, it is possible to define a range for normal traffic in a particular deployment. Thus, when this range is violated, the node could assume that it is under attack. Design of attack countermeasures: Once an attack is detected, a countermeasure must be executed.
These countermeasures should have minimum effect in the normal behavior of the network. Moreover, they should avoid the effects that the attacks produce. With these objectives, the software developers can design the countermeasures and test them in the virtual platform, before network deployment.
These attack countermeasures may use different techniques. The most common methods include turning off attacked nodes, changing the wireless communication channel, changing the encryption key of the communication messages or even excluding the attacker from the network using a filter. The countermeasures are not limited to these methods but they can be as sophisticated as the developer or application requires.
The advantage of the proposed methodology is that these countermeasures can be evaluated and improved before network deployment.
Thus, faults and inefficient implementations can be detected in the early stages of the design process and fixed at low cost. In addition, this methodology allows the comparison of different countermeasures, thus only the most efficient is implemented.
The objective of the framework is to enable the simulation of WSNs in order to analyze the effects of different attacks on the system. The simulator will use the same source code that is executed in the real network nodes.
The WSN virtual platform presented in this paper is based on the native-simulation approach. In this approach, the source code of the WSN nodes is instrumented with additional code that model target-platform WSN mode performance or specific characteristics.
The instrumented code is executed in a host platform desktop computer and it provides estimations during execution.
As is shown in Table 1 , the simulator has some novel features that improve simulation accuracy and facilitate attack simulation. The virtual platform is a software model of the WSN that enables system simulation.
It includes models of the main elements of a WSN: The node model integrates processors, memories, RF-transceivers and sensors see Figure 2. This allows the analysis of the functionality of each WSN node and the estimation its temporal and power consumption behaviors. It also has a reliable network model than can be modified to evaluate any kind of network topology and deployment.
The simulation methodology used in this work is based on the native simulation approach [ 53 ] depicted in Figure 3. SCoPE is used as a starting point of this work that extends this simulator with new capabilities such as WSN and attack simulation. The simulator supports platform modeling for behavioral simulation and performance estimation of embedded systems. The native-simulation based frameworks model all hardware elements related with software execution e.
This approach combines the execution of the annotated software code in a host platform with the use of a virtual platform model of the hardware architecture and embedded RTOS.
With this simulation technique, it is possible to model hardware platform components in System-C and execute the software code of each node on the same platform. Calvo et al.
In Section 4. The co-simulation process includes several steps:. The embedded source code is parsed and analyzed. The basic blocks are identified and annotated with several performance-oriented parameters energy consumption and execution time per basic block, cache and bus access requirements, etc. This compilation process generates an instrumented executable to simulate the system. The execution of this code will produce the performance analysis results.
Additionally, the simulator calculates a Security Estimation Metric during simulation. This metric provides information about the robustness of the encryption that has been used in the WSN transmissions. This metric is presented in [ 14 ]. The network model and the attacker model libraries will be presented in the next subsections. There are two essential components in a wireless sensor node that other systems normally do not integrate. These components are the sensor and the RF transceiver.
The sensor is responsible for collecting external information with a certain period or when an event occurs. This is implemented in the simulation as an external component with specific power consumption and response time.
The sensor model mainly includes the information that has to be transferred to the node, its power consumption, response time and active period. Another important component is the RF transceiver. This is more complex than a sensor and typically integrates a configuration register to control its operation mode.
This paper includes a use-case in which the RF transceiver models an In this case, the implemented registers were:.
Destination Address High and Low: These registers define the message destination address. Baud Rate: Speed for data transfer between transceiver and WSN node controller. Multiple Transmissions: Number of additional broadcast retransmissions. In wireless transmission, the physical channel between two nodes is a shared channel, with limited range, noise and interference. Additionally, the messages can be listened to by other nodes that are not the destinations of the packet.
As a consequence, developers need to determine the node visibility and the probability of a non-successful reception of a packet packet loss probability. The WSN deployment area has to be analyzed and a matrix with the probability of packet loss among all nodes has to be defined. The matrix of packet-loss probability models RF channel characteristics. This probability data may be calculated by the user or by external tools such as an electromagnetic-propagation simulation tool such as Cindoor [ 56 ].
With this matrix, the virtual platform can estimate the connections of the network and the effectiveness of the links between nodes.
It is important to clarify that the network model is responsible for transmitting the packets to their destinations. When a node sends a packet, the network adds the packet to the transmission queue that is sorted be the time of arrival at the reception node.
When the simulation time matches the time of arrival of the packet, the wireless network pops the packet and generates a real random number between 0 and Figure 5 represents a scheme of this wireless network operation. In the network simulator, another important element is the node network interface. This interface is responsible for deciding which packets should actually be received by the node. In a real wireless network, when a node sends a packet to another node, this packet is not only received by the receptor node but also by all the nodes in the transmission range of the sender.
In this case, the node network interface is responsible for disposing of the packets that do not correspond to the node. The network interface checks the packets and the transmission times. In case of package collision two or more packets are overlapped on time , the network interface will discard all the packets involved in the collision.
The interface could also implement the network protocol Zigbee, This allows the modeling of real network transceivers that integrate RF modules with a microcontroller for network protocol management for example, the Xbee [ 55 ]. This facilitates the evaluation of different network protocols.
The previous section presents the infrastructure that allows the simulation and exploration of different WSN design alternatives in normal operation. However, as was mentioned previously, in many cases WSNs are deployed in hostile environments that could put at risk the system behavior. In order to improve network security, it is important to identify the most harmful attacks that a network can suffer. An attack can be defined as an attempt to gain unauthorized access to a service, resource or information.
It could also be an attempt to compromise integrity, availability, or confidentiality of a system. The nature of the attacks is huge enough to make them difficult to classify.
Nevertheless, Mohammadi, S. While passive attacks relate to privacy eavesdropping, gathering and stealing of information by intercepting data communications or monitoring packets exchanged within a WSN active attacks perform actions such as injecting faulty data into the WSN, impersonating, modifying resource and data streams, creating holes in security protocols, destroying sensor nodes, degrading performance, disrupting functionality and overloading the network.
The model and tool presented in this paper are focused on active attacks, which mostly affect network performance. More precisely, this paper addresses those attacks that disrupt, totally or partially, the communication flow among network nodes. Typical WSN attacks can be classified into different categories, according to [ 3 ]. A detailed study of these attacks concludes that they mainly produce two effects. The first effect is the increase in the network traffic due to the introduction of new packets in the network.
The second one is the opposite effect, the reduction of the network traffic due to the elimination or loss of network packets.
These two effects may act together increasing the network traffic of a specific packet and decreasing others. A more detailed study of the attacks shows that not all the attacks can be modeled with these effects, thus an additional special model is required. In summary, WSNs attacks can be classified in four categories depending on their effects on the network:. This section presents a model that groups most of the WSN attacks in four categories, in terms of the attack effects on the network.
The first type of attacks are based on the introduction of fake packets into the network with the aim of making the original nodes process them, increasing the traffic in the network and, thus, congesting it or even disrupting the data of the network. Interrogation attack: Energy Drain [ 58 ]: Due to the difficulty of replacing sensor node batteries and their energy constraints, attackers may use compromised nodes to inject fabricated reports into the network or generate large amounts of traffic in the network.
These fake messages cause false alarms that waste response effort, and drain the finite amount of energy in a battery-powered network.
The aim of this attack is to destroy the sensor nodes in the network, degrade performance of the network and eventually split the network grid up, so taking control of part of the sensor network by inserting a new Sink node. Hello Flood attack [ 59 ]: The attacker typically attempts to drain the energy from a node or exhaust its resources.
Misdirection attack [ 60 ]: The attacker routes the packet from its children to other distant nodes, but not necessarily to its legitimate parent. The main objective of the intruder is to misdirect the incoming messages to increase the latency, which prevents a few packets from reaching the base station.
Flooding attack [ 61 ]: An attacker may repeatedly make new connection requests until the resources required by each connection are exhausted or a maximum limit is reached. It produces severe resource constraints for legitimate nodes. The effects of the attacks placed in this group consist in reducing the traffic in the network.
These attacks are focused on the introduction of noise in the network or other techniques with the objective of increasing the probabilities of packet loss. The main consequence of these attacks is the increment in the packet loss rate which can disrupt the proper function of the network.
The attacks placed in this category are the following:. Jamming attack [ 62 ]: This works by denying service to authorized users as legitimate traffic is jammed by the overwhelming amount of illegitimate traffic. It disrupts network functionality by broadcasting high-energy signals. There are many Jamming attack strategies. Collision attack [ 63 ]: Packets collide when two nodes attempt to transmit on the same frequency simultaneously, producing packet corruption.
This attack can cause a lot of disruption to network operation. Resource Exhaustion attack: Operation of this attack consists in repeated collisions and multiple retransmissions until the node dies. A malicious node continuously requests or transmits over the channel.
Black Hole attack [ 64 ]: A black hole attack basically consists in the network routing alteration with the objective of attracting all the packets to the attacked node destination, and silently discarding or dropping them. Denial of service DoS attacks [ 65 ]: In a Path-based DoS PDoS attack, an adversary swamps sensor nodes a long distance away by flooding a multihop end-to-end communication path with either replicated packets or spurious injected packets.
It can cause serious damage in resource-constrained systems. Homing attack: In a homing attack, the attacker looks at network traffic to deduce the geographic location of critical nodes, such as cluster heads or neighbors of the base station. The attacker can then physically disable these nodes. This leads to another type of black hole attack. Selective Forwarding attack [ 66 ]: Multi-hop networks assume that participating nodes will faithfully forward and receive messages.
However a malicious node may refuse to forward certain messages and simply drop them, ensuring that they are not propagated any further. The procedure to launch selective forward attacks is very similar to the black hole one. First, a malicious node has to convince the network that it is the nearest node to the base station, attracting network traffic to route data through it. Then, a selection of packets is dropped. The third group consists in attacks that cause different effects on the network by mixing effects of the previous ones.
These attacks cause the network to lose some packets, but simultaneously they introduce new packets in the network. Thus, these attacks alter the types of packets transmitted, by reducing the number of some types of packets but increasing others, with the consequent impact on the WSN.
Spoofed attack [ 67 ]: A spoofing attack is a situation in which an attacker successfully masquerades as another node by falsifying data and thereby gaining an illegitimate advantage. This attack consists in targeting routing information while it is being exchanged: Sybil attack [ 68 ]: This attack consists of the modification of the network routing to attract the traffic of the attacker nodes, with the objective of isolating these nodes.
When these nodes can no longer communicate, the attacker sends fake traffic supplanting the nodes. Node replication attack [ 69 , 70 , 71 ]: This is an attack where the attacker tries to mount several nodes with the same identity at different places in the existing network.
Although Sybil attacks and Node Replication attacks might seem similar, these attacks are essentially different. In Sybil attacks, a single node exists with multiple identities while in node replication attacks multiple nodes are present with the same identity. Therefore, in Sybil attack an adversary can succeed by mounting only a single node, whereas a node replication attack requires more nodes to be mounted throughout the network.
In this way, as the number of network nodes increases, the chance to detect this attack also grows. Looping in the network attack: This attack consists of the modification of the network routing by affecting the node data transmission. These attacks usually require direct access to the hardware node tamper attacks. Application attack: It normally requires access to the on-field software update procedure Over the Air Programming procedure or physical access to the hardware of the node.
Overwhelm attack: An attacker might attempt to overwhelm sensor nodes with sensor stimuli that could produce large volumes of traffic to a base station. This induces, among other problems, a power consumption increase in the attacked nodes and the generation of unreliable sensor info. This section defines three attackers that will cover all the attacks presented in the previous subsection. As is mentioned previously, the studied attacks have four different effects. These effects can be modeled with only three attackers.
Therefore, all the mentioned attacks can be modeled with a combination of these three attackers. A typical attack could increase or reduce the network traffic. It can be observed that both attacks are similar: Thus, this type of attacks could be modeled with the same attacker node: This special WSN node is responsible of introducing fake packets in the network.
The structure of this package can be user-defined. The packets are received by the WSN nodes because their structure is formally correct. The simulation of this kind of attackers depends on several parameters that have to be defined during attack configuration. It defines the fake packet rate or number of packets per second that are injected into the network. It defines the type of packets that the attacker injects. Several types of packets have been implemented: It defines the range of time in which the attacker is active.
The attacker can be turned on and turned off many times during the simulation. Nodes Destine: This list specifies the WSN nodes that receive of the injected packets.
If this attribute is defined, each packet will be sent to all nodes. Other attacks could reduce the network traffic. They could be modeled with a new type of attacker: The packets that are affected by these high-energy signals cannot reach their destination the destination node only receives noise and it is not able to decode the packet.
In the case of a black hole attack, the network routing structure is modified with the objective of attracting all the packets to the attacker node. Then the attacker silently discards or drops all the captured packed. As can be appreciated, the effects of both attacks are similar: This attacker is responsible for introducing noise in the network with the objective of increasing the probability of packet loss. This model associates packet-loss probabilities to every possible wireless channel between nodes.
This reduces the communication link quality, thus packets could fail to reach their destination. In order to define a Link-Noise attacker, several parameters can be specified:. List of communication links or node-pairs that are affected by this attacker node. Noise that will be applied to every link that has been defined in the previous parameter. Percentage of packets that will be affected by the increased packet-loss probability.
For other packets, the packet-loss probability will not be affected. It defines the range of time in which the attacker affects the network. The attack will only be active for specific packet types. This enables the simulation of selective attacks. Some attacks could be modeled with a combination of both attackers. Another example of an attack modeled with the combination of both attackers is the node replication attack. It is an attack where the attacker tries to mount several nodes with the same identity at different places in the existing network.
This is the reason why a special attacker is proposed: This attacker will accept two parameters that define the new application code that will be downloaded to the node and the time in which the node will be attacked.
The previous section defines three new attacker models that emulate most of the real WSN attacks. These attacker models are:. Inject packets in the network. Table 2 shows the relation between real WSN attacks and the proposed attacker models. The set of real attacks presented in this section is based on the vulnerabilities described in the state-of-the-art.
These real attacks will be modeled with a proposed attacker or a combination of them. The proposed attacker and the real attack must provide the same effect in the WSN. As can be seen in Table 2 , a Link-Noise node can model different attacks: Different parameter values of the attacker node configuration enable different types of attacks to be modeled.
For example jamming and collision attacks can be modeled with the same attacker node Link-node but with different parameters. These parameters are not only used to define attacks, but also to define attack strategies. The Direct attacker enables the modeling of the Overwhelm and Application attacks. The tampering and sniffing attacks are not modeled because they are passive attacks and they do not affect the operation of the node.
In this case, the tampering attack is based on giving physical access to a node; an attacker can extract sensitive information such as cryptographic keys or other data on the node. As can be appreciated, most WSN attacks can be modeled using the proposed attackers. For this, it is necessary to focus on the effects of each attack and use different attackers to simulate each effect.
Basically, this attacker modifies the packet-loss probability for certain packet types during predefined periods of time. The modification is presented in Figure 6. When a packet has to be transferred to the receiver node, the reception probability will include the original link probability and the additional noise produced by the attacker.
If a Fake packet injector attacker has to be simulated, new packets have to be injected into the transmission queue of the network model shown in Figure 5. Once the packet is inserted in this queue, the network transmits these fake packets as if they were genuine. Figure 7 shows how this attacker modifies the network model in Figure 5.
It models the programming of a genuine node by a fake program. The attack definition includes the new application to be downloaded to the node and its network parameters packet-loss probabilities.
In order to evaluate the proposed attack modeling and simulation techniques, two types of experiments have been performed. The first type of experiments Section 6. The second type Section 6.
This section also demonstrates the advantages of the attack-aware software that was developed with the proposed methodology. In order to demonstrate the suitability of the attack simulation, three network topologies will be evaluated. The three networks have nine nodes with similar hardware architecture and embedded software. These nine-node networks are shown in Figure 8 , Figure 9 and Figure The first one is a meshed wireless network Figure 8.
Figure 9 shows the deployment of a linear wireless network and Figure 10 shows a circular wireless network. The percentages on the red lines represent the packet-loss probabilities of the wireless channel. The simulated networks include two different types of nodes: The Gateway or Central node is responsible for coordinating the network and communicating with other external networks. The Sensor node has a sensor to read the environmental temperature.
When the nodes finish their operation, they change to a sleep mode to reduce power consumption and to increase the battery life. All the simulated nodes integrate a Cortex M4 microcontroller, running at 90 MHz, a memory and an The basic functionality of every node type is briefly commented:.
The Gateway is responsible for receiving messages from the Sensor nodes and transmitting them to an external network. When all the Sensor Nodes are awake, the Gateway waits for all responses with data about environmental temperature. When the Gateway has the information from all of nodes, it composes a message and sends it through a GPRS module. After this, the node sleeps for about 30 s. After waking up, it reads the sensor and sends the data to the Gateway or to another node that can reach the Gateway node.
When it finishes its function, it sleeps for 30 s. Thus, the frequency of data acquisition is every 30 s. Three different attacks have been analyzed. The second attack is a Hello Flow Interrogation attack on the gateway nodes and the third attack consists in a jamming and an injection attack on node 3.
The objective of these simulations is to show how the tool can estimate the attack impact in terms of energy consumption for each node and the network. This section presents the virtual simulator results.
To obtain these results, each network is simulated during one hour. The energy consumption estimations for each node and for the whole network are shown in Table 3 , Table 4 and Table 5. The energy consumption values are shown for the no-attack case. In the case of attack, the tables show the energy increase that the attack produces. It is represented as a percentage: Table 3 shows the results for the linear network. In the case of the linear network, the effect of the Jamming attack is low compared to the other two attacks.
Table 4 shows the results for the Meshed network.
It can be observed that the jamming attack has the highest network impact. Specifically, this attack doubles the energy consumption of the network compared to the no-attack case. It affects all the nodes that compose the network almost equally. For this topology, the Injection attack produces the smallest impact.
Table 5 shows the results for the Circular network. It can be observed that the Jamming and Injection attack have a similar impact on the whole network. Figure 11 shows the absolute values of the total energy consumption for the 12 different cases.
It can be observed that the Meshed network suffers a great impact under a Jamming attack. Thus, embedded software should integrate some countermeasures to reduce the impact of this type of attack.
In addition, it is important to notice that the virtual platform provides estimations that are useful even for no-attack cases.