Artificial Intelligence Enabled Software Defined Networking: A ...

Artificial Intelligence Enabled Software Defined Networking: A Comprehensive Overview Majd Latah, Levent Toker Department of Computer Engineering, Ege University, 35100, Bornova, Izmir, Turkey [email protected], [email protected]

arXiv:1803.06818v1 [cs.AI] 19 Mar 2018

Abstract In recent years, the increased demand for dynamic management of network resources in modern computer networks in general and in today’s data centers in particular has resulted in a new promising architecture, in which a more flexible controlling functionalities can be achieved with high level of abstraction. In software defined networking (SDN) architecture, a central management of the forwarding elements (i.e. switches and routers) is accomplished by a central unit, which can be programmed directly to perform fundamental networking tasks or implementing any other additional services. Combining both central management and network programmability, opens the door to employ more advanced techniques such as artificial intelligence (AI) in order to deal with high-demand and rapidly-changing networks. In this study, we provide a detailed overview of current efforts and recent advancements to include AI in SDN-based networks.

using the SDN paradigm for their own data centers [11, 12]. Artificial intelligence (AI) reveals a huge potential in SDN innovation. Our previous study [13] focused on highlighting the first efforts to integrate AI in SDN. In this work, however, we attempt to we cover the most relevant research efforts to gain deeper insight into the significant role of AI in SDN paradigm.

2

SDN Architecture

As mentioned previously, SDN promotes innovation by introducing the concept of centralized programmable control of the data plane, which facilitates the development of new network services and protocols [4]. The SDN architecture is designed based on the idea of the separation between control and data planes (see Fig. 1). The first attempt was network control point (NCP) [14], which employed the separation concept in order to enhance the control of AT&Ts telephone network, whereas recent contributions such as Ethane [7] and SANE [15] applied the same concept for Ethernet networks [16]. The SDN applications 1 Introduction reside in the application plane of SDN architecture where the Software-defined networking (SDN) adopts the concept of northbound application programming interface (API) provides programmable networks by using a logically centralized man- the commutation between the application and control planes agement, which represents a simplified solution for complex [8], which allows implementing a set of network services such tasks such as traffic engineering [1], network optimization [2] as traffic engineering, intrusion detection, quality of service and orchestration [3]. Furthermore, dealing with modern net- (QoS), firewall and monitoring applications [4]. Northbound work applications requires more scalable architecture which API allows developers to write their own applications without should be able to provide reliable and sufficient services based the need for detailed knowledge of the controller functions or on a specific traffic type [4]. This can be achieved with the understanding how the data plane works. It is worth mentionSDN architecture, which maintains a global view of network ing that several SDN controllers provide their own northbound states and provides a flow-level control of the underlying lay- APIs [16]. ers [4]. This idea caused a dramatical change in the way how The communication between control and data planes is pronetworks are designed and managed [5, 6]. In addition, the vided using a southbound API such as forwarding and conSDN architecture allows the involvement of third parties in trol element separation (ForCES) [17], open vSwitch database the design and deployment of modern network applications (OVSDB) [18], protocol oblivious forwarding (POF) [19], [6]. Ethane project [7] formed the foundation of today’s SDN OpenState [20], OpenFlow (OF) [21] and OpFlex [22], which by presenting a new networking paradigm, in which a central- enables exchanging control messages with forwarding eleized controller is used for flow-level policy management and ments (e.g., OF-enabled switches). As shown in Figure 1, security purposes in enterprise networks. each OF-enabled switch adopts a flow-based decision making The SDN paradigm separates the control plane from the logic determined by the so-called SDN controller, which is redata plane, which results in achieving much faster and dy- sponsible for preparing the forwarding tables of each switch namic approach compared with conventional network archi- [8]. A typical OF-enabled switch has a pipeline of flow tatecture [8]. The control plane can be split into several virtual bles, which consist of flow entries, each of which has three networks where each one implements a different policy [8]. parts: (1) matching rules, which are used for matching inAs a result, this paradigm can be viewed as a tool allows ad- coming packets (2) counters maintain statistics of matched dressing various issues in networking from another perspec- flows and (3) actions or instructions, which can be configured tive [5] and can be used also to fulfill the requirements of new proactively or reactively to be executed upon a match [6, 14]. technologies such as internet of things (IoT) [9]. The adoption The forwarding elements (i.e. OF-enabled switches) can be of SDN paradigm, however, strongly depends on its success implemented in either software or hardware. Some software in reaching an appropriate solution for the problems, which switches such as Open vSwitch have a great potential for procannot be solved by the traditional networking protocols [10]. viding a solution for data centers and virtual networks [16]. Some companies such as Microsoft and Google have started On the other hand, other APIs [10, 23] are proposed for a spe-

Figure 1: A basic SDN architecture cific purpose (e.g, VOIP applications and inter-domain routing, not to mention various SDN programming languages such as Procera [24], NetCore [25] and Frenetic [26], which provide high-level APIs that can be used to write different SDN applications in more flexible and functional manner.

3

hanced matching/action abilities, optical ports, group tables, meter tables and synchronized tables [4, 29]. More details concerning various versions of OF specification can be found in [30]. There are many OF controllers available, such as POX [31], Beacon [32], OpenDayLight [33], Floodlight [34], and Ryu [35].

OpenFlow Protocol 4

OpenFlow (OF) is considered the most commonly used southbound API in SDN, which is being continuously developed and standardized by open networking foundation (ONF) [6]. OF provides an abstraction layer that enables the SDN controller to securely communicate with OF-enabled forwarding elements [6]. OpenFlow has become the de-facto standard for southbound APIs used in SDNs [6] and therefore, in this study, we mainly focus on OF-based SDNs. OF-based forwarding devices have been developed to coexist together with conventional Ethernet devices [16]. Hybrid switches, on the other hand, reveal new possibilities by including both OF and nonOF ports [6]. As we mentioned previously, a set of control messages can be sent by the controller in order to prepare and update switchs flow tables. A typical OF-enabled switch handles new coming packets based on its flow table. Figure 2, shows the fields of matching rules part in OF version 1.0.0. A table-miss occurs when a new packet does not match any of the flow table entries. In this case, the switch may either drop the packet or forward it to the corresponding controller using OF protocol [6]. It is worth to mention that the identity-based access control of the Ethane project [7] became the first specification of OF switches. After the release of the first OF specification in Spring 2008, many vendors such as HP, NEC, Cisco and Juniper were able to build their first OF-enabled hardware switches where NOX [27] was the only available controller at that period of time [28]. Recently, different versions of OF protocol have been introduced in order to add more flexible and reliable capabilities by including multiple flow tables, en-

Artificial Intelligence

Artificial Intelligence (AI) is a rapidly growing field that includes a wide range of subfields, including knowledge representation, reasoning, planning, decision making, optimization, machine learning (ML) and metaheuristic algorithms. Turing Test [36], provided a fulfilling definition of intelligence, where a computer has to answer some questions written by a human interrogator. Accordingly, the computer passes the test if the interrogator cannot tell whether the answer was written by a human or a machine. In order to pass Turing Test, the computer would need to include advanced capabilities such as natural language processing, knowledge representation, automated reasoning, machine learning and computer vision [36]. AI research stared later in the mid-1950s, where a summer workshop organized by Martin Minsky and Claude Shannon at Dartmouth College resulted in the birth of the field of AI [37]. The first contribution, however, was made in 1943 by McCulloch and Pitts, proposing the first model for artificial neural networks, in which each neuron has a binary output (-1,+1) with a sign activation function [37]. Adoption of AI approaches increased, thanks to key contributions led to the emergence of new subfields such as expert systems, fuzzy logic and evolutionary computation. Further efforts have fueled the research in AI field by refining the existing methods and proposing brand new hybrid intelligent approaches. Machine learning, metaheuristic algorithms and fuzzy inference systems are widely used in SDN paradigm therefore we provide an introduction regarding these approaches. 2

Figure 2: Structure of OpenFlow V1.0.0 matching rules

4.1

Machine Learning

An intelligent machine learns from experience (i.e., learns from the data available in its environment) and uses it to improve the overall performance [36, 37]. In this context, learning methods fall into four types: 4.1.1

Supervised learning

Supervised learning methods are provided with a pre-defined knowledge. For instance, a training dataset that consists of input-output pairs, where the system learns a function that maps a given input to an appropriate output [36]. This approach requires having a dataset that represents the system under consideration and can be used to estimate the performance of the selected method [38]. 4.1.2

Unsupervised learning

Unsupervised learning methods are provided without a predefined knowledge (i.e., having an unlabeled data) [36]. Therefore, the system mainly focuses on finding specific patterns in the input. An example of the unsupervised learning approach is clustering, which is used for detecting useful clusters in the input data based on similar properties defined by a proper distance metric such as Euclidian, Jaccard, and cosine distance metrics [36, 38]. 4.1.3

Reinforcement learning

In reinforcement learning approach, the system learns based on a set of reinforcements from its environment. For instance, a reward or punishment determines whether the system performed well or not [36]. 4.1.4

Semi-supervised learning

In semi-supervised learning the system learns from both labeled and unlabeled data, where the lack of labels as well as the labeled part may contain a random noise forms a situation between supervised and unsupervised learning [36].

4.2

solve difficult problems such as combinatorial, highly nonlinear and multimodal optimization problems [44]. Each one of these algorithms has different advantages, therefore many efforts were made to design hybrid approaches that combine their benefits and ultimately attain better results [44]. The success of these algorithms is determined by achieving a balanced approach between the exploration and the exploitation [39]. Exploitation is useful for determining the most promising high-quality solutions in the search space, whereas exploitation is needed to concentrate search in some areas based on previous search results [39].

4.3

Fuzzy Inference Systems

Unlike classical binary logic where any fact can only be true or false, fuzzy logic (FL) is a multi-valued logic, which deals with a degree of truth or degree of membership (i.e., any value between 0 and 1) [37]. Therefore, Boolean logic can be seen as a special case of FL. Fuzzy systems (FS) are used for mapping a given input to an appropriate output based on the principles of fuzzy set theory, which was developed by Lotfi Zaheh [37, 46]. FS make use of fuzzy rules in the form: If x is A and y is B then z is k

(1)

where x, y and z are linguistic variables; A and B are linguistic values determined by fuzzy sets on the universe of discourses X, Y respectively and k is a constant that represents the consequent of the rule. As shown in Fig. 3, the system converts the crisp input to the appropriate fuzzy sets using the membership functions. Thereafter, it will be evaluated using the inference engine. Finally, the output is determined using an appropriate defuzzification method such as center of gravity (COG) or weighted average (WA) [37]. Mamdani and Sugeno methods are two inference techniques that can be used in fuzzy systems. Unlike Mamdani’s approach which employs a fuzzy membership function of the rule consequent, Sugeno’s approach uses a single point to represent the rule consequent. Sugenos approach, is considered to be more efficient than Mamdani’s approach which requires finding the centroid of a two-dimensional shape and this in turn increases the computational cost [37].

Metaheuristic Algorithms

The heuristic algorithms use a problem-specific heuristic whereas the metaheuristics form an efficient general purpose approach that includes a wide range of application area ranging from finance to engineering and networking [39]. Metaheuristics have been increasingly employed to solve hard optimization problems [40], that cannot be solved by any deterministic (exact) approach within a reasonable time [39]. Most of these algorithms are nature-inspired [40], from simulated annealing (SA) [41] to genetic algorithms (GA) [42], and from ant colony optimization (ACO) [43] to whale optimization algorithm (WOA) [44]. These algorithms are widely used to

5

Artificial Intelligence in SDN

AI and ML approaches have been widely used for solving various problems such as routing [47], traffic classification [48], flow clustering [49] intrusion detection [50], load balancing [51], fault detection [52], quality of service (QoS) and quality of experience (QoE) optimization [53], admission control and resource allocation [54]. However, in the era of SDN the role of AI was significantly increased due to the huge efforts made by industry and research community. Recent studies have shown a strong tendency of research community towards

Figure 3: A structure of a typical fuzzy inference system adoption of AI approaches in SDNs. It is worth mentioning that ML, metaheuristics and fuzzy systems were the most common approaches for solving various networking related problems. We investigate the application of each approach in SDN paradigm separately as will be shown in the reset of this study.

5.1

Machine Learning in SDN

5.1.1

Supervised learning in SDN

In this context, neural networks [55-63], support vector machine [65-71], random forest [72-78], decision trees [79-92] and deep learning [93-94] approaches were the most used supervised learning techniques in SDN. From Table 1, we observe that neural networks (NN) approach was used mainly for intrusion detection and prevention [55, 57-58], solving controller placement problem [57], load balancing [59], performance perdition [60], service level agreements (SLA) enforcement [61, 62], routing and optimal virtual machine (VM) placement [63]. Table 2, on the other hand, shows that support vector machine (SVM) approach was mainly used for deploying intrusion detection systems in the SDN paradigm [65-71]. As shown in Table 3, the random forest (RF) approach was widely used for estimating service-level metrics [72], intrusion detection, security applications [73-74], QoE prediction [75], traffic classification [76] and threat assessment [77, 78]. Decision trees were widely used for application identification [79, 80], intrusion detection [81] as well as solving SDNrelated security challenged [82]. Furthermore, deep learning techniques have shown promising results in SDN compared to traditional ML approaches as shown in Table 5. 5.1.2

Unsupervised learning in SDN

K-means [95-99] and self-organizing map [100-104] were the most used unsupervised learning techniques in SDN paradigm. Research efforts made to apply k-means approach to SDNs are summarized in Table 6. On the other hand, self-organizing map (SOM) approach was used widely in the SDN paradigm for intrusion detection as shown in Table 7. Other unsupervised ML methods such as hidden markov model (HMM) and restricted Boltzmann machine were used in the SDN paradigm for security assessment and intrusion detection (see Table 8). 5.1.3

Reinforcement learning in SDN

In this context, it is worth noting that Q-learning technique [110-114] was applied widely in SDN paradigm for routing

and adaptive video streaming. The application of reinforcement approach to SDN is summarized in Table 9. 5.1.4

Semi-supervised learning in SDN

Semi-supervised learning [115-117] were also used in SDN but much less common compared with other learning approaches. As shown in Table 10, the research was focused on traffic classification [115, 117] and routing [116].

5.2

Metaheuristic Algorithms Used in SDN Paradigm

A large variety of metaheuristic algorithms such as ant colony optimization [118-126], genetic algorithms [127-133], evolutionary algorithms [134-137], particle swarm optimization [138-142] and simulated annealing [143-145] were widely adopted in SDN. Research efforts made to apply metaheuristic approach to SDN are summarized in Table 11.

5.3

Fuzzy Inference Systems in SDN

Fuzzy inference systems were also widely used in SDN paradigm [152-157]. The main research was focused on introducing new protocols [152], intrusion detection [153-155], selection the optimal network deployment schemes [156] and traffic engineering [157].

5.4

Other AI-based Approaches in SDN

Other researchers proposed many hybrid models that make use of two or more intelligent approach in order to enhance the overall performance and accuracy of these algorithms as shown in Table 13.

6

Conclusion

In this paper, we provided a state-of-the-art overview of research efforts made for applying artificial intelligence techniques in SDN paradigm. Compared to our recent paper [13], this study showed an increased adoption of various AI techniques to solve a wide range of networking problems and address new challenges introduced by the SDN paradigm. This study showed that ML, metaheuristics and fuzzy systems were the most common AI approaches used in SDN. Many other hybrid AI approaches were used for achiever higher accuracy compared to other traditional AI approaches. We conclude that AI techniques form a significant component for future networks in general and SDN paradigm in particular.

Ref [55] [56]

[57] [58] [59] [60]

[61, 62]

[63]

Brief Summary A distributed intrusion prevention system that can be used in large scale networks. The system achieved lower computational and communication overhead due to its parallel and simple computational capabilities. A multi-label classification to predict global network allocations (i.e. weighted controller placement problem). Compared to decision tree and logistic regression, neural network approach showed better results and saved up to two-thirds of the algorithm runtime. A neural network was used for offline prediction of traffic demands in a mobile network operator reducing the optimality gap below 0.2% (virtual wavelength path-hourly-NN) and 0.45% (wavelength path-hourly-NN). In addition, NN approach was used for offline prediction of the next configuration time point. An intrusion detection system for SDN based on NN approach which achieved high accuracy reached 97.3% accuracy using NSL-KDD dataset. NN approach for load balancing. Compared to [64] and static Round Robin strategy, the experimental study showed that this method can achieve better performance and resulted in 19.3% decreasing of network latency. A NN approach based on Levenberg Marquardt (LM) algorithm to predict the performance of SDN according to round-trip time (RTT) and throughput. A new approach to enforce service level agreements (SLA) in SDN and virtualized network functions (NFV). The research focused on prediction of service level objectives (SLOs) breaches for streaming services running on NFV and SDN. The research showed that long short term memory (LSTM) is more robust than feed forward neural networks. A new paradigm that employs AI in SDN, termed as knowledge-defined networking (KDN). The paradigm included a proof of concept concerning routing in an overlay network based on NN approach where MSE reached 1%. In addition to solving the problem of optimal virtual machine (VM) placement in NFV paradigm.

Table 1: The application of neural network approach to the SDN paradigm

Ref [65] [66] [67, 68] [69] [70] [71]

Brief Summary Detection of DDoS attacks on the SDN controller. SVM showed higher accuracy and less false positive rate compared to other classifiers. Detection of flooding attacks based on SVM and idletimeout adjustment (IA) algorithm which leads to an accurate system and reduction of CPU usage of Open vSwitches and their correspondence SDN controller. An enhanced behavior-based SVM approach for network intrusion [3] and malware [4] detection where information gain (IG) metric was used for selecting the most relevant features. A new approach that combines SVM with self-organizing map (SOM) in order to get a higher accuracy and better detection rate for DDoS attacks in SDN as well as achieving lower false alarm rate. Detection of DDoS TCP flood attacks based on least squares support vector machine (LS-SVM), which outperformed the following algorithms: naive bayes, k-nearest neighbor and multilayer perceptron (MLP). An SVM approach for implementing a traffic-based filtering classifier, which represents the second stage of a framework proposed in order to mitigate DoS attacks.

Table 2: The application of SVM approach to the SDN paradigm

Ref [72] [73] [74] [75] [76] [77] [78]

Brief Summary An RF approach for estimating service-level metrics based on network-level statistics. Intrusion detection system where the RF approach has been used for both feature selection and classification. An SDN-based system for automatic identification and security enforcement of potentially vulnerable IoT devices using device type-specific profiling approach. ML based QoE prediction in SDN paradigm. The authors utilized the following ML algorithms: RF, k-nearest neighbors (k-NN), NN and decision trees where RF showed better performance. Several ML techniques such as random forest, stochastic gradient boosting and extreme gradient boosting were employed for traffic classification. An RF approach showed the best performance among k-NN, naive bayes and logistic regression (LR) for cyber threat detection. An approach for analyzing, detecting and reacting against cyber threats. Experimental resulted based on ISOT botnet dataset showed that RF approach achieves the best performance among k-NN, naive bayes and logistic regression.

Table 3: The application of RF approach to the SDN paradigm

5

Ref

Brief Summary The C4.5 decision tree approach was used for application identification in order to associate each application type with a corresponding QoS level. An intrusion detection and prevention system based on C4.5 algorithm. Employing a decision tree approach in mitigation of host location hijacking attacks on SDN controllers. A framework for detection and prevention of simple mail transfer protocol (SMTP) flow attacks. Flow based botnet detection based on C4.5 algorithm in SDN. Traffic classification for mobile agents in SDN based on C5.0 algorithm. Solving flow table congestion problem based on C4.5 algorithm. Prediction of potential vulnerable hosts using historical network data and ML algorithms such as C4.5, bayesian network (BN), decision table (DT), and naive-bayes (NB). The best results were achieved by bayesian network approach. Introducing new extensions for decision-tree algorithms, namely adaptive grouping factor (AGF) and independent sub-rule (ISR) leaf structure achieving better performance for larger rules as well as reducing the number of memory access by a factor of 3. This study considers SDN rule-sets. A two-stage approach for detection of elephant flows where the first stage consists of an efficient sampling and the next stage, employs C4.5 for correlated flows. Prediction of QoS violations based on M5Rules appoarch which combines decision trees and linear regression. Integrating a J48-tree classifier for intrusion detection on OpenFlow switches. Detecting Botnets using SDN paradigm, where decision trees showed better results for detecting peer-to-peer botnets whereas SVM and Bayesian networks were more effective in detecting command and control (C&C) related botnets such as HTTP and IRC (internet relay chat) botnets.

[79, 80] [81] [82] [83] [84] [85] [86] [87] [88] [89] [90] [91] [92]

Table 4: The application of decision trees approach to the SDN paradigm

Ref [93] [94]

Brief Summary Intrusion detection system based on a simple deep neural network, where deep learning approach achieved the best results compared to other supervised ML algorithms such as naive bayes, SVM and decision trees. Construction of routing table based on supervised deep belief networks. The experimental results showed that the proposed approach outperformed traditional open shortest path first (OSPF) protocol in terms of throughout and average delay per hope.

Table 5: The application of deep learning approach to the SDN paradigm

Ref [95] [96] [97] [98] [99]

Brief Summary K-means and probabilistic transition based approach were used for detecting application layer DDoS attacks. SDN-based approach for achieving Wi-Fi direct clustering in campus networks. Solving the problem of optimal controller placement based on k-means approach. K-means was used for user traffic profiling in campus SDN. Naive bayes, k-NN, k-means and k-medoids were used for detecting DDoS attacks in SDN, where the best results achieved by naive bayes approach.

Table 6: The application of k-means approach to the SDN paradigm

Ref [100] [101] [102] [103] [104]

Brief Summary Intrusion detection based on SOM approach. The main disadvantage in this approach is high-grained flow-based traffic matching and sampling. Detecting and mitigating DoS attacks based on a combination of access control and SOM classification. The proposed method can deal with IP/MAC spoofing attacks. Distributed SOM approach for tackling the performance bottleneck and overload problems for large-sized SDNs under flooding attacks. Detecting DDoS flooding attacks based on SOM approach. Comparing different ML approaches for intrusion detection in SDN paradigm. The experimental study showed that hierarchical learning vector quantization (LVQ1) is more efficient than other approaches such as SOM and LVQ1.

Table 7: The application of SOM approach to the SDN paradigm

Ref [105] [106] [107] [108]

Brief Summary Anomaly detection and mitigation based on a two-stage unsupervised learning approach for feature selection and clustering. Mitigation of domain name system (DNS) query-based DDoS attacks based on Dirichlet process mixture model for clustering traffic flows. The proposed system achieved better results than mean shift (MS) clustering approach. Security situation assessment for SDN based on hidden markov model (HMM) where Baum-Welch algorithm was used for training and Viterbi algorithm was used for predicting the network state. Detection of DDoS attacks in SDN based on restricted boltzmann machine.

Table 8: The application of other unsupervised ML approaches to the SDN paradigm

6

[109] [110] [111] [112]

[113]

[114]

Routing based on reinforcement learning approach which showed less loss rate and better jitter values, when compared to traditional OSPF routing protocol. QoS-aware adaptive routing (QAR) for distributed hierarchical control planes where markov decision processes (MDP) with QoS-aware reward function used for modeling the system. Instead of using conventional Q-learning method, softmax action selection policy and stateactionrewardstateaction (SARSA) approach were used for quality update. Congestion prevention based on Q-learning approach. Compared with Dijkstra’s algorithm and extended Dijkstra’s algorithm the proposed approach showed better results when the size of transmitted data increases. Adaptive video streaming based on Q-learning approach where markov decision process (MDP) was used for modeling the system which showed better results when compared with the shortest path routing and greedy-based approaches. An end-to-end SDN-based intelligent architecture for large-scale HTTP adaptive streaming (HAS) systems where partially observable markov decision process (POMDP) was used for modeling the system and Q-learning based approach that employs a per-cluster decision algorithm to maximize QoE. The experimental study showed that the system was able to increase the video stability and achieve better QoE fairness and network resource utilization. An online Q-learning-based dynamic bandwidth allocation algorithm (Q-FDBA) for better QoE fairness. Q-FDBA showed better results when compared to bandwidth-aware (BA) streaming and QoE fairness framework (QFF).

Table 9: The application of reinforcement learning approaches to the SDN paradigm

[115]

[116]

[117]

Proposed a new framework for QoS-aware traffic classification based on semi-supervised learning. This approach can classify the network traffic according to the QoS requirements. The system allows achieving deep packet inspection (DPI) and semi-supervised learning based on Laplacian SVM. The Laplacian SVM approach outperformed a previous semi-supervised approach based on k-means classifier. Introduced an efficient routing pre-design solution based on semi-supervised approach. The study suggested using an appropriate clustering algorithm such as Gaussian mixture model and k-means clustering for feature extraction. Thereafter, a supervised classification approach such as extreme learning machine (ELM) can be used for flow demand forecasting. The authors also suggested using an adaptive multipath routing approach based on analytic hierarchy process (AHP) for handling to elephant flows according different constraint factor weights. Proposed a new method for fine-grained traffic classification based on semi-supervised approach called nearest application based cluster classifier (NACC). Unlike traditional methods which use one feature vectors, this algorithm constructs a matrix with several cluster centroids based on k-means clustering to represent the application. The algorithm uses a small number of labeled flows to build a supervised dataset. Then collects the unlabeled flows to be merged with previously collected dataset, based on investigating the correlated flows, which is used in order to map an application to different clusters. The experimental study showed a good identification accuracy reaching 90%.

Table 10: The application of semi-supervised learning approaches to the SDN paradigm Ref Ant colony optimization (ACO)

Genetic algorithms (GA) Evolutionary algorithm (EA) Particle swarm optimization(PSO) Simulated annealing (SA) Artificial bee colony (ABC) Honeybee foraging algorithm (HFA) Binary bat algorithm (BBA) PSO and firefly algorithm (FA) Whale optimization algorithm (WOA) Chaotic grey wolf optimization algorithm (CGWOA)

Brief Summary Used for solving various networking problems such as routing [118- 120], load balancing [121, 122], video surveillance [123], network security [124, 125] and maximizing network utilization [126]. Used for routing [127], virtual network planning [128], optimizing the cost of deploying deep packet inspection functions in SDN [129], load balancing [130-132] and designing optimal observation matrices [133]. Used for solving controller placement problem in distributed SDNs [134], routing [135], moving target defense [136] and designing optimal observation matrix for network measurement [137]. Used for routing [138, 139], resource management [140], multi-tenant virtual network customization [141], controller provision and switch assignment [142] for cloud datacenters Used for solving controller placement problem [143, 144] and dynamic controller provisioning [145]. Used for traffic engineering for SDN-based video surveillance systems [146]. Used for achieving load balancing for SDN-based cloud systems [147]. Used for feature selection for anomaly detection [148]. Used for solving controller placement problem in SDN-based wide area networks [149]. Used for achieving clustering for heterogeneous, randomly distributed and dense IoT networks [150]. Solving controller placement problem in software-defined mobile networks [151].

Table 11: The application of metaheuristic algorithms to the SDN paradigm Ref [152] [153] [154] [155] [156] [157]

Brief Summary Topology discovery protocol for software defined wireless sensor networks. The proposed protocol showed better performance when compared with software defined networking solution for wireless sensor networks (SDN-WISE) [158]. Prevention of DDoS attacks based on Sugeno’s fuzzy inference approach. A cooperative defense approach against DoS flood attacks based on fuzzy modeling. An information security management system based on Mamdani’s fuzzy inference approach. Threat degree is determined by the fuzzy inference module which mainly depends on threshold random walk with credit-based rate limiting (TRW-CB) and rate limiting algorithms. Selection of network deployment schemes based on Mamdani’s approach. Traffic,engineering method for SDN-based video surveillance systems using adaptive,type-2 fuzzy approach which showed better results when compared with type-1,fuzzy approach, available bandwidth based traffic engineering approach and,OSPF routing protocol in terms of end to end delay, peak signal to noise,ratio (PSNR) and packet loss ratio.

Table 12: The application of fuzzy inference approaches to the SDN paradigm 7

8

[166]

[165]

[164]

[163]

[162]

[161]

[160]

[159]

Ref

Table 13: The application of AI-based hybrid approach to the SDN paradigm

Brief Summary Preventing video freezes in HTTP adaptive streaming based on combination of random undersampling boosting (RUSBoost) algorithm and fuzzy logic approach. The Experimental results showed that the proposed system was able to reduce video freezes with 65% and freeze time with 45%, when compared with fair in-network enhanced adaptive streaming (FINEAS) and Microsoft ISS smooth streaming (MSS). Mitigating DoS/DDoS attacks based on a game theatrical approach that employs Holt-Winters and genetic algorithm with fuzzy logic. Holt-Winters for digital signature (HWDS) approach was used for detecting and identifying anomalies and game theory decision-making (GTDM) approach was used for choosing the optimal defense strategy. HWDS was compared with fuzzy genetic algorithm digital signature (Fuzzy-GADS) approach where HWDS fared better in performance tests. Fuzzy-GADS, on the other hand, found to be more efficient on detecting the occurrence of stealthier anomalies. An intelligent hybrid intrusion detection system for SDN-based 5G network where random forest approach was used for feature selection and k-means++ with Adaboost approach was used for flow classification. Anomaly detection and classification based on two phases: 1) lightweight phase based on entropy analysis and 2) heavyweight approach based on a hybrid ML approach which employs k-means and SVM algorithms. Optimization the performance of SDN-based on hybrid intelligence approach which makes use of a neural network model. GA and PSO algorithms were used separately to select the optimal set of inputs that maximizes the network efficiency. The experimental results showed that PSO achieves a better performance and a faster convergence compared with GA approach. An intelligent SDN-based video management system which uses face detection and recognition techniques based on EigenFace algorithm principal component analysis (PCA), FisherFace algorithm based on linear discriminant analysis (LDA), and local binary patterns (LBP). Revenue maximization in data centers based on workload-aware SDN controller which determines the optimal combination of a VM and routing path for each application. The proposed system uses a hybrid approach named hybrid chaotic simulated-annealing PSO (HCSP) that combines chaotic PSO and SA algorithm. Compared with OSPF and Round Robin approaches, the proposed approach was successful in reducing the RTT and increasing the revenue. Application identification based on DNS responses inspection and ML. A classification system, which adopts a voting strategy, was implemented by Weak Java API from which three algorithms were used namely random forest, rotation forest and random committee with random tree. The experimental results showed that the combination of DNS responses inspection and ML techniques was able to achieve higher accuracy compared to standalone ML techniques.

References [1] S. Raza, G. Huang, C. N. Chuah, S. Seetharaman, J. P. Singh. Measurouting: a framework for routing assisted traffic monitoring. IEEE/ACM Transactions on Networking. 2012; 20(1):45-56. DOI: 10.1109/TNET.2011.2159991 [2] V. Heorhiadi, M. K. Reiter, V. Sekar. Simplifying Software-Defined Network Optimization Using SOL. In: Proceedings of the 13th Usenix Conference on Networked Systems Design and Implementation; 16-18 March 2016; California. Berkeley: USENIX Association; 2006. p. 223-237.

[13] M. Latah, L. Toker. Application of Artificial Intelligence to Software Defined Networking: A Survey. Indian Journal of Science and Technology, 2016;9(44):1-7. DOI: 10.17485/ijst/2016/v9i44/89812 [14] D. Sheinbein, R. P. Weber. Stored program controlled network: 800 service using spc network capability. Bell Labs Technical Journal. 1982;61(7):17371744. DOI: 10.1002/j.1538-7305.1982.tb04370.x [15] M. Casado, T. Garfinkel, A. Akella, M. J. Freedman, D. Boneh, N. McKeown, et al. SANE: A protection architecture for enterprise networks. in Proc. 15th Conf. USENIX Security Symp; 31 July-4 August 2006; Vancouver. Berkeley: USENIX Association; 2006. p. 1-15.

[3] B. Martini, D. Adami, A. Sgambelluri, M. Gharbaoui, L. Donatini, S. Giordano, et al. An SDN orchestrator for resources chaining in cloud data centers. In: Proceedings of 2014 European Conference on Networks and Communications (EuCNC); 23-26 June 2014; Bologna. New York: IEEE; 2014. p. 1-5.

[16] D. Kreutz, F. M. Ramos, P. E. Verissimo, C. E. Rothenberg, S. Azodolmolky, S. Uhlig. Softwaredefined networking: A comprehensive survey. Proceedings of the IEEE. 2015;103(1):14-76. DOI: 10.1109/JPROC.2014.2371999

[4] I. F. Akyildiz, A. Lee, P. Wang, M. Luo, W. Chou. A roadmap for traffic engineering in SDN-OpenFlow networks. Computer Networks. 2014;71:1-30. DOI: 10.1016/j.comnet.2014.06.002

[17] Internet engineering task force (IETF). Forwarding and Control Element Separation (ForCES) Protocol Specification [Internet]. 2010. Available from: http://www.ietf.org/rfc/rfc5810.txt [Accessed: 2017-125]

[5] N. Feamster, J. Rexford, E. Zegura. The road to SDN: an intellectual history of programmable networks. ACM SIGCOMM Computer Communication Review. 2014;44(2):87-98. DOI: 10.1145/2602204.2602219

[18] IETF. The Open vSwitch Database Management Protocol [Internet]. 2013. Available from: https://tools.ietf.org/html/rfc7047 [Accessed: 2017-125]

[6] B. A. A. Nunes, M. Mendonca, X. N. Nguyen, K. Obraczka, T. Turletti. A survey of softwaredefined networking: Past, present, and future of programmable networks. IEEE Communications Surveys & Tutorials. 2014;16(3):1617-1634. DOI: 10.1109/SURV.2014.012214.00180

[19] H. Song. Protocol-oblivious forwarding: Unleash the power of SDN through a future-proof forwarding plane. In: Proc. 2nd ACM SIGCOMM Workshop Hot Topics Softw. Defined Netw.; 16-16 August 2013; Hong Kong. New York: ACM; 2013. p. 127132.

[7] M. Casado, M. J. Freedman, J. Pettit, J. Luo, N. McKeown, S. Shenker. Ethane: Taking control of the enterprise. ACM SIGCOMM Computer Communication Review. 2007;37(4):112. DOI: 10.1145/1282427.1282382

[20] G. Bianchi, M. Bonola, A. Capone, C. Cascone. OpenState: programming platform-independent stateful openflow applications inside the switch. ACM SIGCOMM Computer Communication Review. 2014;44(2):44-51. DOI: 10.1145/2602204.2602211

[8] R. Jain, S. Paul. Network virtualization and software defined networking for cloud computing: a survey. IEEE Communications Magazine. 2013;51(11):24-31. DOI: 10.1109/MCOM.2013.6658648 [9] S. Bera, S. Misra, A. V. Vasilakos. Software-Defined Networking for Internet of Things: A Survey. IEEE Internet of Things Journal. 2017;4(6):1994-2008. DOI: 10.1109/JIOT.2017.2746186

[21] N. McKeown, T. Anderson, H. Balakrishnan, G. Parulkar, L. Peterson, J. Rexford J, et al. OpenFlow: enabling innovation in campus networks. ACM SIGCOMM Computer Communication Review. 2008;38(2):69-74. DOI: 10.1145/1355734.1355746 [22] IETF. OpFlex Control Protocol [Internet]. 2014. Available from: http://tools.ietf.org/html/draft-smith-opflex00. [Accessed: 2017-10-16]

[10] P. Lin, J. Bi, S. Wolff, Y. Wang, A. Xu, Z. Chen, H. Hu, Y. Lin. A west-east bridge based SDN inter-domain testbed. IEEE Communications Magazine. 2015;53(2):190-197. DOI: 10.1109/MCOM.2015.7045408

[23] A. D. Ferguson, A. Guha, C. Liang, R. Fonseca, S. Krishnamurthi. Participatory networking: An API for application control of SDNs. ACM SIGCOMM computer communication review. 2013;43(4):327-338. DOI: 10.1145/2534169.2486003

[11] S. Jain, A. Kumar, S. Mandal, J. Ong, L. Poutievski, A. Singh A, et al. ACM SIGCOMM Computer Communication Review. 2013;43(4):3-14. DOI: 10.1145/2534169.2486019

[24] A. Voellmy, H. Kim, N. Feamster. 2012, August. Procera: a language for high-level reactive network control. In: Proceedings of the first workshop on Hot topics in software defined networks; 13-13 August 2012; Helsinki. New York: ACM; 2012. p. 4348.

[12] C. Hong, S. Kandula, R. Mahajan, M. Zhang, V. Gill, M. Nanduri, et al. Achieving high utilization with softwaredriven WAN. In: proceedings of ACM SIGCOMM conference; 12-16 August 2013; Hong Kong. New York: ACM; 2013. p. 15-26.

[25] C. Monsanto, N. Foster, R. Harrison, D. Walker. A compiler and run-time system for network programming languages. ACM SIGPLAN Notices. 2012;47: 217-230. DOI: 10.1145/2103621.2103685

[26] N. Foster, R. Harrison, M. J. Freedman, C. Monsanto, J. Rexford, A. Story, et al. Frenetic: A network programming language. In ACM Sigplan Notices. 2011;46(9):279-291. DOI: 10.1145/2034574.2034812 [27] NOX An OpenFlow Controller [Internet]. 2008. Available from: http://www.noxrepo.org [Accessed: 2017-1210] M. Kobayashi, S. Seetharaman, G. Parulkar, G. Appenzeller, J. Little, J. Van Reijendam, et al. Maturing of OpenFlow and software-defined networking through deployments. Computer Networks. 2014;61:151-175. DOI: 10.1016/j.bjp.2013.10.011 [28] T. Dargahi, A. Caponi, M. Ambrosin, G. Bianchi, M. Conti. A Survey on the Security of Stateful SDN Data Planes. IEEE Communications Surveys & Tutorials. 2017;19(3):1701-1725. DOI: 10.1109/COMST.2017.2689819 [29] ONF. OpenFlow Specification [Internet]. 2009. Available from: https://www.opennetworking.org/softwaredefined-standards/specifications/ [Accessed: 2017-1212] [30] POX Wiki [Internet]. 2011. Available from: https://openflow.stanford.edu/display/ONL/POX+Wiki [Accessed: 2017-12-12] [31] Beacon [Internet]. 2010. Available from: https://openflow.stanford.edu/display/Beacon/Home [Accessed: 2017-12-12] [32] OpenDayLight [Internet]. 2014. Available from: http://www.opendaylight.org/ [Accessed: 2017-12-12] [33] Floodlight [Internet]. 2012. Available from: http://www.projectfloodlight.org/floodlight/ [Accessed: 2017-12-12] [34] Ryu [Internet]. 2014. Available from: http://osrg.github.com/ryu/ [Accessed: 2017-12-12] [35] S. Russell, P. Norvig. Artificial Intelligence (A Modern Approach). 3rd ed. New Jersey: Prentice Hall; 1995. 1152 p. [36] M. Negnevitsky. Artificial Intelligence - A Guide to Intelligent Systems. 2nd ed. Essex: Addison-Wesley; 2005. 415 p. [37] T. T. Nguyen, G. Armitage. A survey of techniques for internet traffic classification using machine learning. IEEE Communications Surveys & Tutorials. 2008;10(4):56-76. DOI: 10.1109/SURV.2008.080406 [38] I. BoussaD, J. Lepagnot, P. Siarry. A survey on optimization metaheuristics. Information Sciences, 2013;237:82117. DOI: 10.1016/j.ins.2013.02.041 [39] X. S. Yang. Metaheuristic optimization: algorithm analysis and open problems. Lecture Notes in Computer Science. 2011;6630:21-32. DOI: 10.1007/978-3-64220662-7 2 [40] S. Kirkpatrick, C. D. Gellat, M. P. Vecchi. Optimization by simulated annealing. Science 1983;220(4598):670680. DOI: 10.1126/science.220.4598.671

[41] J. H. Holland. Adaption in Natural and Artificial Systems. Ann Arbor: University of Michigan Press; 1975. 183 p. [42] M. Dorigo, M. Birattari, T. Stutzle. Ant colony optimization. IEEE computational intelligence magazine, 2006;1(4):28-39. DOI: 10.1109/MCI.2006.329691 [43] S. Mirjalili, A. Lewis. The whale optimization algorithm. Advances in Engineering Software. 2016;95:5167. DOI: 10.1016/j.advengsoft.2016.01.008 [44] S. Fong, X. Wang, Q. Xu, R. Wong, J. Fiaidhi, S. Mohammed. Recent advances in metaheuristic algorithms: Does the Makara dragon exist?. The Journal of Supercomputing. 2016;72(10):3764-3786. DOI: 10.1007/s11227-015-1592-8 [45] L. A. Zadeh. Fuzzy sets. Information and Control. 1965;8(3):338353. DOI: 10.1016/S00199958(65)90241-X [46] T. Gui, C. Ma, F. Wang, D. E. Wilkins. Survey on swarm intelligence based routing protocols for wireless sensor networks: An extensive study. In: Proc. of IEEE International Conference on Industrial Technology (ICIT); 1417 March 2016; Taipei. New York: IEEE; 2016. p. 19441949. [47] M. Soysal, E. G. Schmidt. Machine learning algorithms for accurate flow-based network traffic classification: Evaluation and comparison. Performance Evaluation. 2010;67(6):451-467. DOI: 10.1016/j.peva.2010.01.001 [48] A. McGregor, M. Hall, P. Lorier, J. Brunskill. Flow clustering using machine learning techniques. In: Proc. of International Workshop on Passive and Active Network Measurement (PAM); 19-20 April 2014; Pins. Berlin: Springer; 2004. p.205-214. [49] X. Xu, X. Wang. An adaptive network intrusion detection method based on PCA and support vector machines. In: Proc. of Advanced Data Mining and Applications; 22-24 July 2005; Wuhan. Berlin: Springer; 2005. p. 696703. [50] H. Y. Kim, J. M. Kim. A load balancing scheme based on deep-learning in IoT. Cluster Computing. 2017;20(1):873-878. DOI: 10.1007/s10586-016-0667-5 [51] A. I. Moustapha, R. R. Selmic. Wireless sensor network modeling using modified recurrent neural networks: Application to fault detection. IEEE Transactions on Instrumentation and Measurement. 2008;57(5):981-988. DOI: 10.1109/TIM.2007.913803 [52] M. S. Mushtaq, B. Augustin, A. Mellouk. Empirical study based on machine learning approach to assess the QoS/QoE correlation. In: Proc. of 17th European Conference on Networks and Optical Communications (NOC); 20-22 June 2012; Vilanova i la Geltru. New York: IEEE; 2016. p. 1-7. [53] A. Testolin, M. Zanforlin, M.D. F. De Grazia, D. Munaretto, A. Zanella, M. Zorzi, et al. A machine learning approach to QoE-based video admission control and resource allocation in wireless systems. In: Proc. of 13th Annual Mediterranean Ad Hoc Networking Workshop (MED-HOC-NET); 2-4 June 2014; Piran. New York: IEEE; 2014. p. 31-38.

[54] X. F. Chen, S. Z. Yu. CIPA: A collaborative intrusion prevention architecture for programmable network and SDN. Computers and Security. 2016;58:119. DOI: 10.1016/j.cose.2015.11.008 [55] M. He, P. Kalmbach, A. Blenk, W. Kellerer, S. Schmid. Algorithm-Data Driven Optimization of Adaptive Communication Networks. In Proc. of IEEE 25th International Conference on Network Protocols (ICNP); 10-13 Oct. 2017; Toronto. New York: IEEE; 2017. p. 1-6. [56] R. Alvizu, S. Troia, G. Maier, A. Pattavina. Matheuristic with machine-learning-based prediction for softwaredefined mobile metro-core networks. IEEE/OSA Journal of Optical Communications and Networking, 2017;9(9):D19-D30. DOI: 10.1364/JOCN.9.000D19

[66] P. Wang, K. M. Chao, H. C. Lin, W. H. Lin, C. C. Lo. An Efficient Flow Control Approach for SDN-Based Network Threat Detection and Migration Using Support Vector Machine. In Proc. of IEEE 13th International Conference on e-Business Engineering (ICEBE); 4-6 Nov. 2016; Macau. New York: IEEE; 2017. p. 56-63. [67] L. Boero, M. Marchese, S. Zappatore. Support Vector Machine Meets Software Defined Networking in IDS Domain. In Proc. of IEEE 29th International Teletraffic Congress (ITC 29); 4-8 Sept. 2017; Genoa. New York: IEEE; 2017. p. 25-30.

[57] A. Abubakar, B. Pranggono. Machine learning based intrusion detection system for software defined networks. In Proc. of Seventh International Conference on Emerging Security Technologies (EST); 6-8 Sept. 2017; Canterbury. New York: IEEE; 2017. p. 138-143.

[68] T. V. Phan, N. K. Bao, M. Park. A Novel Hybrid Flowbased Handler with DDoS Attacks in Software-Defined Networking. In: Proc. of 2016 Intl IEEE Conferences on Ubiquitous Intelligence & Computing, Advanced and Trusted Computing, Scalable Computing and Communications, Cloud and Big Data Computing, Internet of People, and Smart World Congress; 18-21 July 2016; Toulouse. New York: IEEE; 2016. p. 350-357.

[58] C. Chen-Xiao, X. Ya-Bin. Research on load balance method in SDN. International Journal of Grid and Distributed Computing. 2016;9(1):25-36. DOI: 10.14257/ijgdc.2016.9.1.03

[69] A. Sahi, D. Lai, Y. Li, M. Diykh. An Efficient DDoS TCP Flood Attack Detection and Prevention System in a Cloud Environment. IEEE Access. 2017;5:6036-6048. DOI: 10.1109/ACCESS.2017.2688460

[59] A. Sabbeh, Y. Al-Dunainawi, H. S. Al-Raweshidy, M. F. Abbod. Performance prediction of software defined network using an artificial neural network. In Proc. of SAI Computing Conference (SAI); 13-15 July 2016; London. New York: IEEE; 2017. p. 80-84.

[70] G. Shang, P. Zhe, X. Bin, H. Aiqun, R. Kui. FloodDefender: Protecting data and control plane resources under SDN-aimed DoS attacks. In Proc. of IEEE Conference on Computer Communications (INFOCOM 2017); 1-4 May 2017; Atlanta. New York: IEEE; 2017. p. 1-9.

[60] J. Bendriss, I. G. B. Yahia, P. Chemouil, D. Zeghlache. AI for SLA Management in Programmable Networks. In Proc. of 13th International Conference on Design of Reliable Communication Networks (DRCN); 8-10 March 2017; Munich. Berlin: VDE; 2017. p. 1-8.

[71] R. Stadler, R. Pasquini, V. Fodor. Learning from Network Device Statistics. Journal of Network and Systems Management. 2017;25(4):672-698. DOI: 10.1007/s10922-017-9426-z

[61] J. Bendriss, I. G. B. Yahia, D. Zeghlache. Forecasting and anticipating SLO breaches in programmable networks. In Proc. of 20th Conference on Innovations in Clouds, Internet and Networks (ICIN); 7-9 March 2017. Paris. New York: IEEE; 2017. p. 127-134.

[72] C. Song, Y. Park, K. Golani, Y. Kim, K. Bhatt, K. Goswami. Machine-Learning Based Threat-Aware System in Software Defined Networks. In Proc. of 26th International Conference on Computer Communication and Networks (ICCCN); 13 July-3 Aug. 2017; Vancouver. New York: IEEE; 2017. p. 1-9.

[62] A. Mestres, A. Rodriguez-Natal, J. Carner, P. Barlet-Ros, E. Alarcn, M. Sol, et al. Knowledge-defined networking. ACM SIGCOMM Computer Communication Review. 2017;47(3):2-10. DOI: 10.1145/3138808.3138810 [63] Y. Li, D. Pan. OpenFlow based load balancing for Fat-Tree networks with multipath support. In Proc. 12th IEEE International Conference on Communications (ICC13); 9-13 June. 2013; Budapest. New York: IEEE; 2013. p. 1-5. [64] R. T. Kokila, S. T. Selvi, K. Govindarajan. DDoS detection and analysis in SDN-based environment using support vector machine classifier. In Proc. of IEEE Sixth International Conference on Advanced Computing (ICoAC); 17-19 Dec. 2014; Chennai. New York: IEEE; 2015. p. 205-210. [65] T. V. Phan, T. Van Toan, D. Van Tuyen, T. T. Huong, N. H. Thanh. OpenFlowSIA: An optimized protection scheme for software-defined networks from flooding attacks. In Proc. of IEEE Sixth International Conference on Communications and Electronics (ICCE); 27-29 July 2016; Ha Long. New York: IEEE; 2016. p. 13-18.

[73] M. Miettinen, S. Marchal, I. Hafeez, N. Asokan, A. R. Sadeghi, S. Tarkoma. IoT Sentinel: Automated devicetype identification for security enforcement in IoT. In Proc. of IEEE 37th International Conference on Distributed Computing Systems (ICDCS); 5-8 June 2017; Atlanta. New York: IEEE; 2017. p. 2177-2184. [74] T. Abar, A. B. Letaifa, S. El Asmi. Machine learning based QoE prediction in SDN networks. In Proc of. 13th International Wireless Communications and Mobile Computing Conference (IWCMC); 26-30 June 2017; Valencia. New York: IEEE; 2017. p. 1395-1400. [75] P. Amaral, J. Dinis, P. Pinto, L. Bernardo, J. Tavares, H. S. Mamede. Machine Learning in Software Defined Networks: Data collection and traffic classification. In Proc. of IEEE 24th International Conference on Network Protocols (ICNP); 8-11 Nov. 2016; Singapore. New York: IEEE; 2017. p. 1-5. [76] M. Zago, V. M. Ruiz Snchez, M. G. Prez, G. M. Prez. Tackling cyber threats with automatic decisions and reactions based on machine-learning techniques. In Proc.

of the 2nd Conference on Network Management, Quality of Service and Security for 5G Networks; 12-15 June 2016; Oulu. p. 1-4. [77] G. A. Ajaeiya, N. Adalian, I. H. Elhajj, A. Kayssi, A. Chehab. Flow-based Intrusion Detection System for SDN. In Proc. of 2017 IEEE Symposium on Computers and Communications (ISCC); 3-6 July 2017; Heraklion. New York: IEEE; 2017. p. 787-793. [78] F. Li, J. Cao, X. Wang, Y. Sun. A QoS Guaranteed Technique for Cloud Applications Based on Software Defined Networking. IEEE Access. 2017;5:21229-21241. DOI: 10.1109/ACCESS.2017.2755768 [79] S. T. V. Pasca, S. S. P. Kodali, K. Kataoka. AMPS: Application aware multipath flow routing using machine learning in SDN. In Proc. of IEEE Twenty-third National Conference on Communications (NCC); 2-4 Mar. 2017; Chennai. New York: IEEE; 2017. p. 1-6. [80] A. Le, P. Dinh, H. Le, N. C. Tran. Flexible NetworkBased Intrusion Detection and Prevention System on Software-Defined Networks. In Proc. of IEEE International Conference on Advanced Computing and Applications (ACOMP); 23-25 Nov. 2015; Ho Chi Minh. New York: IEEE; 2016. p. 106-111. [81] R. Nagarathna, S. M. Shalinie. SLAMHHA: A supervised learning approach to mitigate host location hijacking attack on SDN controllers. In IEEE Fourth International Conference on Signal Processing, Communication and Networking (ICSCN); 16-18 Mar. 2017; Chennai. New York: IEEE; 2017. p. 1-7. [82] M. Z. A. Aziz, K. Okamura. A Method to Detect SMTP Flood Attacks using FlowIDS Framework. International Journal of Computer Science and Network Security, 2017;17(6):14-21. [83] F. Tariq, S. Baig. Botnet classification using centralized collection of network flow counters in software defined networks. International Journal of Computer Science and Information Security, 2016;14(8):1075-1080. [84] Z. A. Qazi, J. Lee, T. Jin, G. Bellala, M. Arndt, G. Noubir. Application-awareness in SDN. ACM SIGCOMM computer communication review. 2013;43(4):487-488. DOI: 10.1145/2534169.2491700 [85] B. Leng, L. Huang, C. Qiao, H. Xu. A decision-treebased on-line flow table compressing method in Software Defined Networks. In Proc. of IEEE/ACM 24th International Symposium Quality of Service (IWQoS); 2021 June 2016; Beijing. New York: IEEE; 2017. p. 1-2. [86] S. Nanda, F. Zafari, C. DeCusatis, E. Wedaa, B. Yang. Predicting network attack patterns in SDN using machine learning approach. In Proc. of IEEE Conference on Network Function Virtualization and Software Defined Networks (NFV-SDN); 7-10 Nov. 2016; Palo Alto. New York: IEEE; 2017. p. 167-172. [87] T. Stimpfling, N. Blanger, O. Cherkaoui, A. Bliveau, L. Bliveau, Y. Savaria. Extensions to decision-tree based packet classification algorithms to address new classification paradigms. Computer Networks. 2017;122:83-95. DOI: 10.1016/j.comnet.2017.04.021

[88] F. Tang, L. Li, L. Barolli, C. Tang. An Efficient Sampling and Classification Approach for Flow Detection in SDNBased Big Data Centers. In Proc. of IEEE 31st International Conference on Advanced Information Networking and Applications (AINA); 27-29 Mar. 2017; Taipei. New York: IEEE; 2017. p. 1106-1115. [89] S. Jain S, M. Khandelwal, A. Katkar, J. Nygate. Applying big data technologies to manage QoS in an SDN. In Proc. of 12th International Conference on Network and Service Management (CNSM); 31 Oct.-4 Nov. 2016; Montreal. New York: IEEE; 2017. p. 302-306. [90] N. T. Van, H. Bao, T. N. Thinh. An Anomaly-based Intrusion Detection Architecture Integrated on OpenFlow Switch. In Proceedings of the 6th International Conference on Communication and Network Security; 26-29 Nov. 2016; Singapore. New York: ACM; 2016. p. 99103. [91] U. Wijesinghe, U. Tupakula, V. Varadharajan. Botnet detection using software defined networking. In Proc. of IEEE 22nd International Conference on Telecommunications (ICT); 27-29 April. 2015; Sydney. New York: IEEE; 2015. p. 219-224. [92] T. A. Tang, L. Mhamdi, D. McLernon, S. A. R. Zaidi , M. Ghogho. Deep Learning Approach for Network Intrusion Detection in Software Defined Networking. In Proc. of IEEE International Conference on Wireless Networks and Mobile Communications (WINCOM); 26-29 Oct. 2016; Fez. New York: IEEE; 2017. p. 258-263. [93] B. Mao, Z. M. Fadlullah, F. Tang, N. Kato, O. Akashi, T. Inoue, et al. Routing or Computing? The Paradigm Shift Towards Intelligent Computer Network Packet Transmission Based on Deep Learning. IEEE Transactions on Computers. 2017;66(11):1946:1960. DOI: 10.1109/TC.2017.2709742 [94] E. Ivannikova, M. Zolotukhin, T. Hmlinen. Probabilistic Transition-Based Approach for Detecting ApplicationLayer DDoS Attacks in Encrypted Software-Defined Networks. In International Conference on Network and System Security; 21-23 Aug. 2017; Helsinki. Cham: Springer; 2017. p. 531-543. [95] T. M. T. Nguyen, L. Hamidouche, F. Mathieu, S. Monnet, S. Iskounen. SDN-based Wi-Fi Direct clustering for cloud access in campus networks. Annals of Telecommunications. DOI: 10.1007/s12243-017-0598-z [96] K. S. Sahoo, S. Sahoo, S. K. Mishra, S. Mohanty, B. Sahoo. Analyzing Controller Placement in Software Defined Networks. In: IJCA Proceedings on National Conference on Next Generation Computing and its Applications in Computer Science and Technology NGCAST ; 27-28 Sept. 2016; Sarang. IJCA Journal; 2017. p. 16-12. [97] T. Bakhshi, B. Ghita. OpenFlow-enabled user traffic profiling in campus software defined networks. In Proc. of IEEE 12th International Conference on Wireless and Mobile Computing, Networking and Communications (WiMob); 17-19 Oct. 2016; New York. New York: IEEE; 2016. p. 1-8. [98] L. Barki, A. Shidling, N. Meti, D. G. Narayan, M. M. Mulla. Detection of distributed denial of service attacks

in software defined networks. In IEEE International Conference on Advances in Computing, Communications and Informatics (ICACCI); 21-24 Sept. 2016; Jaipur. New York: IEEE; 2016. p. 2576-2581. [99] D. Jankowski, M. Amanowicz. Intrusion detection in software defined networks with self-organized maps. Journal of Telecommunications and Information Technology. 2015;4:39. [100] T. Wang, H. Chen. SGuard: A lightweight SDN safe-guard architecture for DoS attacks. China Communications. 2017;14(6):113-125. DOI: 10.1109/CC.2017.7961368 [101] T. V. Phan, N. K. Bao, M. Park. Distributed-SOM: A novel performance bottleneck handler for large-sized software-defined networks under flooding attacks. Journal of Network and Computer Applications. 2017;91:1425. DOI: 10.1016/j.jnca.2017.04.016 [102] R. Braga, E. Mota, A. Passito. Lightweight DDoS flooding attack detection using NOX/OpenFlow. In Proc. of IEEE 35th Conference on Local Computer Networks (LCN); 10-14 Oct. 2010; Denver. New York: IEEE; 2011. p. 408-415. [103] D. Jankowski, M. Amanowicz. On Efficiency of Selected Machine Learning Algorithms for Intrusion Detection in Software Defined Networks. International Journal of Electronics and Telecommunications. 2016;62(3):247-252. DOI: 10.1515/eletel-2016-0033 [104] D. He, S. Chan, X. Ni, M. Guizani. SoftwareDened-Networking-Enabled Trafc Anomaly Detection and Mitigation. IEEE Internet of Things Journal. 2017;4(6):1890:1898. DOI: 10.1109/JIOT.2017.2694702 [105] M. E. Ahmed, H. Kim, M. Park. Mitigating DNS Query-Based DDoS Attacks with Machine Learning on Software-Defined Networking. In Proc. of The 36th IEEE Military Communications Conference; 23-25 Oct. 2017; Baltimore. New York: IEEE; 2017. p. 1-6. [106] Z. Fan, Y. Xiao, A. Nayak, C. Tan. An improved network security situation assessment approach in software defined networks. Peer-to-Peer Networking and Applications. DOI: 10.1007/s12083-017-0604-2 [107] P. MohanaPriya, S. M. Shalinie. Restricted Boltzmann Machine based detection system for DDoS attack in Software Defined Networks. In Proc. of IEEE Fourth International Conference on Signal Processing, Communication and Networking (ICSCN); 16-18 Mar. 2017; Chennai. New York: IEEE; 2017. p. 1-6.

[110] S. Kim, J. Son, A. Talukder, C. S. Hong. Congestion prevention mechanism based on Q-leaning for efficient routing in SDN. In Proc. of International Conference on Information Networking (ICOIN); 13-15 Jan. 2016; Kota Kinabalu. New York: IEEE; 2016. p. 124-128. [111] T. Uzakgider, C. Cetinkaya, M. Sayit. Learning-based approach for layered adaptive video streaming over SDN. Computer Networks. 2015;92(P2):35768. DOI: 10.1016/j.comnet.2015.09.027 [112] A. Bentaleb, A. C. Begen, R. Zimmermann, S. Harous. SDNHAS: An SDN-Enabled architecture to optimize QoE in http adaptive streaming. IEEE Transactions on Multimedia. 2017;19(10):2136-2151. DOI: 10.1109/TMM.2017.2733344 [113] J. Jiang, L. Hu, P. Hao, R. Sun, J. Hu, H. Li. Q-FDBA: improving QoE fairness for video streaming. Multimedia Tools and Applications. DOI: 10.1007/s11042-0174917-1 [114] P. Wang, S. C. Lin, M. Luo. A framework for QoSaware traffic classification using semi-supervised machine learning in SDNs. In Proc. of IEEE International Conference on Services Computing (SCC); 27 June-2 July 2016; San Francisco. New York: IEEE; 2016. p. 760-765. [115] F. Chen, X. Zheng. Machine-learning based routing pre-plan for sdn. In International Workshop on Multidisciplinary Trends in Artificial Intelligence; 13-15 Nov 2015; Fuzhou. Cham: Springer; 2015. p. 149-159. [116] W. Li, G. Li, X. Yu. A fast traffic classification method based on SDN network. In: Proc. of the 4th International Conference on Electronics, Communications and Networks (CECNET IV); 12-15 Dec. 2014; Beijing. London: CRC Press; 2015. p. 223-229. [117] O. Dobrijevic, M. Santl, M. Matijasevic. Ant colony optimization for QoE-centric flow routing in softwaredefined networks. In Proc. of 11th International Conference on Network and Service Management (CNSM); 9-13 Nov. 2015; Barcelona. New York: IEEE; 2016. p. 274-278. [118] F. Wang, B. Liu, L. Zhang, Q. Zhang, Q. Tian, F. Tian, et al. Dynamic routing and spectrum assignment based on multilayer virtual topology and ant colony optimization in elastic software-defined optical networks. Optical Engineering. 2017;56(7):076111. DOI: 10.1117/1.OE.56.7.076111

[108] S. Sendra, A. Rego, J. Lloret, J. M. Jimenez, O. Romero. Including artificial intelligence in a routing protocol using Software Defined Networks. In Proc. of IEEE International Conference on Communications Workshops (ICC Workshops); 21-25 May 2017; Paris. New York: IEEE; 2017. p. 670-674.

[119] C. Gao, H. Wang, L. Zhai, S. Yi, X. Yao. Optimizing Routing Rules Space through Traffic Engineering Based on Ant Colony Algorithm in Software Defined Network. In Proc. of IEEE 28th International Conference on Tools with Artificial Intelligence (ICTAI); 6-8 Nov. 2016; San Jose. New York: IEEE; 2017. p. 106-112.

[109] S. C. Lin, I. F. Akyildiz, P. Wang, M. Luo. QoSaware adaptive routing in multi-layer hierarchical software defined networks: a reinforcement learning approach. In Proc. of IEEE International Conference on Services Computing (SCC); 27 June-2 July 2016; San Francisco. New York: IEEE; 2016. p. 25-33.

[120] A. Di Stefano, G. Cammarata, G. Morana, D. Zito. A4sdn-adaptive alienated ant algorithm for softwaredefined networking. In Proc. of 10th International Conference on P2P, Parallel, Grid, Cloud and Internet Computing (3PGCIC); 4-6 Nov. 2015; Krakow. New York: IEEE; 2016. p. 344-350.

[121] C. Wang, G. Zhang, H. Xu, H. Chen. An ACO-based Link Load-Balancing Algorithm in SDN. In: Proc. of 7th International Conference on Cloud Computing and Big Data (CCBD); 16-18 Nov. 2016; Macau. New York: IEEE; 2016. p. 214-218.

[132] M. Malboubi, Y. Gong, Z. Yang, X. Wang, C. N. Chuah, P. Sharma. Software defined network inference with evolutionary optimal observation matrices. Computer Networks. 2017;129:93-104. DOI: 10.1016/j.comnet.2017.09.001

[122] M. R. Parsaei, R. Mohammadi, R. Javidan. A new adaptive traffic engineering method for telesurgery using ACO algorithm over Software Defined Networks. European Research in Telemedicine/La Recherche Europenne en Tlmdecine. 2017;6(3-4):173:180. DOI: 10.1016/j.eurtel.2017.10.003

[133] T. Zhang, P. Giaccone, A. Bianco, S. De Domenico. The role of the inter-controller consensus in the placement of distributed SDN controllers. Computer Communications. 2017;113:1-13. DOI: 10.1016/j.comcom.2017.09.007

[123] H. H. Chen, S. K. Huang. LDDoS Attack Detection by Using Ant Colony Optimization Algorithms. Journal of Information Science & Engineering. 2016;32(4):9951020. [124] J. Liu, H. Zhang, Z. Guo. A Defense Mechanism of Random Routing Mutation in SDN. IEICE TRANSACTIONS on Information and Systems. 2017;100(5):10461054. DOI: 10.1587/transinf.2016EDP7377 [125] S. Yi, H. Wang, X. Yao, C. Gao, L. Zhai. Maximizing Network Utilization for SDN Based on WiseAnt Colony Optimization. In Proc. of IEEE 18th International Conference on High Performance Computing and Communications; IEEE 14th International Conference on Smart City; IEEE 2nd International Conference on Data Science and Systems (HPCC/SmartCity/DSS); 12-14 Dec. 2016; Sydney. New York: IEEE; 2017. p. 959-966. [126] Y. S. Yu, C. H. Ke. Genetic algorithmbased routing method for enhanced video delivery over software defined networks. International Journal of Communication Systems. 2017;31(1):e3391. DOI: 10.1002/dac.3391 [127] J. Wang, C. De Laat, Z. Zhao. QoS-aware virtual SDN network planning. In IFIP/IEEE Symposium on Integrated Network and Service Management (IM); 8-12 May 2017; Lisbon. New York: IEEE; 2017. p. 644-647. [128] M. Bouet, J. Leguay, V. Conan. Cost-based placement of virtualized deep packet inspection functions in sdn. In Proc. of IEEE Military Communications Conference; 18-20 Nov. 2013; San Diego. New York: IEEE; 2013. p. 992-997.

[134] A. Fernndez-Fernndez, C. Cervell-Pastor, L. OchoaAday. A Multi-Objective Routing Strategy for QoS and Energy Awareness in Software-Defined Networks. IEEE Communications Letters. 2017;21(11):2416-2419. DOI: 10.1109/LCOMM.2017.2741944 [135] A. Makanju, A. N. Zincir-Heywood, S. Kiyomoto. On evolutionary computation for moving target defense in software defined networks. In: Proc. of the Genetic and Evolutionary Computation Conference Companion; 1519 July 2017; Berlin. New York: ACM; 2017. p. 287288. [136] M. Malboubi, Y. Gong, W. Xiong, C. N. Chuah, P. Sharma. Software defined network inference with passive/active evolutionary-optimal probing (sniper). In Proc. of 24th International Conference on Computer Communication and Networks (ICCCN); 3-6 Aug. 2015; Las Vegas. New York: IEEE; 2015. p. 1-8. [137] M. K. Awad, M. ElShafei, T. Dimitriou, Y. Rafique, M. Baidas, A. Alhusaini. Powerefficient routing for SDN with discrete link rates and sizelimited flow tables: A treebased particle swarm optimization approach. International Journal of Network Management. 2017;27(5): e1972. DOI: 10.1002/nem.1972 [138] Y. Xiong, J. Shi, Y. Lv, G. N. Rouskas. Power-Aware Lightpath Management for SDN-Based Elastic Optical Networks. In Proc. of 26th International Conference on Computer Communication and Networks (ICCCN); 31 July-3 Aug. 2017; Vancouver. New York: IEEE; 2017. p. 1-9. [139] Z. X. Chang, H. Zhou, W. J. Yang. Optimization of Resource Management for 5G. DEStech Transactions on Computer Science and Engineering. 2017;538-543. DOI: 10.12783/dtcse/cece2017/14594

[129] L. D. Chou, Y. T. Yang, Y. M. Hong, J. K. Hu, B. Jean. A genetic-based load balancing algorithm in openflow network. In: Huang YM, Chao HC, Deng DJ, Park J, editors. Advanced Technologies, Embedded and Multimedia for Human-centric Computing. Dordrecht: Springer; 2014. p. 411-417. DOI: 10.1007/978-94-007-7262-5 48

[140] K. Li, J. Bao, Z. Lu, Q. Qi, J. Wang. A PSO-based virtual SDN customization for multi-tenant cloud services. In Proceedings of the 11th International Conference on Ubiquitous Information Management and Communication; 5-7 Jan. 2017; Beppu. New York: ACM; 2017. p. 1-6.

[130] S. B. Kang, G. I. Kwon. Load Balancing Strategy of SDN Controller Based on Genetic Algorithm. Advanced Science and Technology Letters. 2016;129:219222. DOI: 10.14257/astl.2016.129.43

[141] A. Farshin, S. Sharifian. MAP-SDN: a metaheuristic assignment and provisioning SDN framework for cloud datacenters. The Journal of Supercomputing, 2017;73(9):41124136. DOI: 10.1007/s11227-017-20012

[131] U. Mahlab, P. E. Omiyi, H. Hundert, Y. Wolbrum, O. Elimelech, I. Aharon, et al. Entropy-based loadbalancing for software-defined elastic optical networks. In Proc. of 19th International Conference on Transparent Optical Networks (ICTON); 2-6 July 2017; Girona. New York: IEEE; 2017. p. 1-4.

[142] Y. Hu, W. Wendong, X. Gong, X. Que, C. Shiduan. Reliability-aware controller placement for softwaredefined networks. In IFIP/IEEE International Symposium on Integrated Network Management (IM 2013); 27-31 May 2013; Ghent. New York: IEEE; 2013. p. 672675.

[143] H. Selvi, S. Gner, G. Gr, F. Alagz. A genetic-based load balancing algorithm in openflow network. In: Liyanage

Conference on Fuzzy Systems (FUZZ-IEEE); 9-12 July 2017; Naples. New York: IEEE; 2017. p. 1-8.

[144] M, Gurtov A, Ylianttila M, editors. Software Defined Mobile Networks (SDMN): Beyond LTE Network Architecture. Chichester: Wiley; 2015. p. 129-147. DOI: 10.1002/9781118900253.ch8

[155] S. Dotcenko, A. Vladyko, I. Letenko. A fuzzy logicbased information security management for softwaredefined networks. In Proc. of 16th International Conference on Advanced Communication Technology (ICACT); 16-19 Feb. 2014; Pyeongchang. New York: IEEE; 2014. p. 167-171.

[145] M. F. Bari, A. R. Roy, S. R. Chowdhury, Q. Zhang, M. F. Zhani, R. Ahmed, R. Boutaba. Dynamic controller provisioning in software defined networks. In Proc. of 9th International Conference on Network and Service Management (CNSM); 14-18 Oct. 2013; Zurich. New York: IEEE; 2014. p. 18-25. [146] R. Mohammadi, R. Javidan. An Intelligent Traffic Engineering Method over Software Defined Networks for Video Surveillance Systems Based on Artificial Bee Colony. International Journal of Intelligent Information Technologies (IJIIT). 2016;12(4):4562. DOI:10.4018/IJIIT.2016100103 [147] B. Kang, H. Choo. An SDN-enhanced load-balancing technique in the cloud system. The Journal of Supercomputing. DOI: https://doi.org/10.1007/s11227-016-1936-z [148] R. Sathya, R. Thangarajan. Efficient anomaly detection and mitigation in software defined networking environment. In Proc. of 2nd International Conference on Electronics and Communication Systems (ICECS); 2627 Feb. 2015; Coimbatore. New York: IEEE; 2015. p. 479-484. [149] K. S. Sahoo, A. Sarkar, S. K. Mishra, B. Sahoo, D. Puthal, M. S. Obaidat, et al. Metaheuristic Solutions for Solving Controller Placement Problem in SDN-based WAN Architecture. In: Proc. of 8th International Conference on Data Communication Networking; 26-28 July 2017. Madrid. Setubal: SciTePress; 2017. p. 15-23. [150] T. A. Al-Janabi, H. S. Al-Raweshidy. Efficient whale optimisation algorithm-based SDN clustering for IoT focused on node density. In Proc. of 16th Annual Mediterranean Ad Hoc Networking Workshop (Med-Hoc-Net); 28-30 June 2017; Budva. New York: IEEE; 2017. p. 1-6.

[156] S. Abdallah, I. H. Elhajj, A. Chehab, A. Kayssi. Fuzzy decision system for technology choice in hybrid networks. In Proc. of Fourth International Conference on Software Defined Systems (SDS); 8-11 May 2017; Valencia. New York: IEEE; 2017. p. 106-111. [157] R. Mohammadi, R. Javidan. An adaptive type-2 fuzzy traffic engineering method for video surveillance systems over software defined networks. Multimedia Tools and Applications. 2016;76(22):23627-23642. DOI: 10.1007/s11042-016-4137-0 [158] L. Galluccio, S. Milardo, G. Morabito, S. Palazzo. SDN-wise: design, prototyping and experimentation of a stateful SDN solution for wireless sensor networks. In Proc. of 2015 IEEE Conference on Computer Communications (INFOCOM); pp. 513521. [159] S. Petrangeli, T. Wu, T. Wauters, R. Huysegems, T. Bostoen, F. De Turck. A machine learning-based framework for preventing video freezes in HTTP adaptive streaming. Journal of Network and Computer Applications. 2017;94:78-92. DOI: 10.1016/j.jnca.2017.07.009 [160] M. V. De Assis, A. H. Hamamoto, T. Abrao, M. L. Proenca. A Game Theoretical Based System Using Holt-Winters and Genetic Algorithm with Fuzzy Logic for DoS/DDoS Mitigation on SDN Networks. IEEE Access. 2017;5:9485-9496. DOI: 10.1109/ACCESS.2017.2702341 [161] J. Li, Z. Zhao, R. Li. A Machine Learning Based Intrusion Detection System for Software Defined 5G Network. IET Networks. DOI: 10.1049/iet-net.2017.0212

[151] A. Farshin, S. Sharifian. A chaotic grey wolf controller allocator for Software Defined Mobile Network (SDMN) for 5th generation of cloud-based cellular systems (5G). Computer Communications. 2017;108:94109. DOI: 10.1016/j.comcom.2017.05.003

[162] A. S. Da Silva, J. A. Wickboldt, L. Z. Granville, A. Schaeffer-Filho. ATLANTIC: A framework for anomaly traffic detection, classification, and mitigation in SDN. In Proc. of 2016 IEEE/IFIP Network Operations and Management Symposium (NOMS); 26 April-1 May 2015; Kowloon. New York: IEEE; 2015. p. 27-35.

[152] N. Abdolmaleki, M. Ahmadi, H. T. Malazi, S. Milardo. Fuzzy topology discovery protocol for SDNbased wireless sensor networks. Simulation Modelling Practice and Theory. 2017;79:54-68. DOI: 10.1016/j.simpat.2017.09.004

[163] A. Sabih, Y. Al-Dunainawi, H. S. Al-Raweshidy, M. F. Abbod. Optimisation of Software-Defined Networks Performance Using a Hybrid Intelligent System. Advances in Science, Technology and Engineering Systems. 2017;2(3):617-622. DOI: 10.25046/aj020379

[153] P. Van Trung, T. T. Huong, D. Van Tuyen, D. M. Duc, N. H. Thanh, A. Marshall. A multi-criteria-based DDoSattack prevention solution using software defined networking. In Proc. of International Conference on Advanced Technologies for Communications (ATC); 14-16 Oct. 2015; Ho Chi Minh. New York: IEEE; 2016. p. 308313.

[164] S. Li, Z. Guo, Y. Liu, G. Shou, Y. Hu. The intelligent video management system: A use case of software defined class. In Proc. of 12th International Conference on Computer Science and Education (ICCSE); 25-29 April 2016; Istanbul. New York: IEEE; 2016. p. 149-154.

[154] A. A. Rezaei, R. Mohammadifar, E. Lotfi, A. Khosravi, S. Nahavandi. A heterogeneous defense method using fuzzy decision making. In Proc. of IEEE International

[165] H. Yuan, J. Bi, J. Zhang, W. Tan, K. Huang. WorkloadAware Revenue Maximization in SDN-Enabled Data Center. In IEEE 10th International Conference on Cloud Computing (CLOUD), 25-30 June 2017; Honolulu. New York: IEEE; 2017. p. 18-25.

[166] N. F. Huang, C. C. Li, C. H. Li, C. C. Chen, C. H. Chen, I. H. Hsu. Application identification system for SDN QoS based on machine learning and DNS responses. In Proc. of 19th Asia-Pacific Network Operations and Management Symposium (APNOMS); 7-29 Sept. 2017; Seoul. New York: IEEE; 2017. p. 407-410.