SDN with Internet of Things

IoT with SDN. Software Defined Network (SDN) provides the flexibility in traditional networking by decoupling Data plane (hardware) and Control plane(software). SDN helps in virtualization of resources, network management, and development of applications without changing underlying hardware. The focus is, on application side with centralized management of network. On the other side, aim of Internet of Things(IoT) is to connect billions of objects with internet. Our study focus is to see the benefits for IoT from emerging these two different technologies. We conduct a comparative study from different perspective to see how SDN can adapt the current system in support of different IoT challenges including security management, routing, adaptability, heterogeneity, scalability, and network cost. We see the challenges overcome by SDN for IoT. Also, see to what extent flexibility is provided by SDN to control and manage the billions of IoT devices efficiently

Emergence of SDN with IoT Muhammad Kareem Punjab University College of Information Technology Phdcsf18m501@pucit.edu.pk Abstract Software Defined Network (SDN) provides the flexibility in traditional networking by decoupling Data plane (hardware) and Control plane(software) SDN helps in virtualization of resources, network management, and development of applications without changing underlying hardware The focus is, on application side with centralized management of network On the other side, aim of Internet of Things(IoT) is to connect billions of objects with internet Our study focus is to see the benefits for IoT from emerging these two different technologies We conduct a comparative study from different perspective to see how SDN can adapt the current system in support of different IoT challenges including security management, routing, adaptability, heterogeneity, scalability, and network cost We see the challenges overcome by SDN for IoT Also, see to what extent flexibility is provided by SDN to control and manage the billions of IoT devices efficiently Keywords: Software Defined Networking (SDN), Internet of Things(IoT), Load Balancing (LB), 6LowPAN, Wireless Sensor Networks(WSN’s) 1.Introduction Internet of things has taken much attraction of researchers in recent years It can manage objects remotely, efficiently, and quickly IoT is capable to connect anything at anyplace and anytime with anyone [1] IoT applications are contributing in every field of life including home automation, smart cities, livestock and logistics, transport management and smart grids Aim of IoT is to connect billions of objects with internet This is difficult to manage huge and heterogenous network of IoT On the other side, Software Defined Networking (SDN) has made new paradigm in network management SDN provides opportunities to solve rigidity of traditional network In traditional networking data plane and control plane were combined in each device including routers, load balancer and switches SDN come up with the solution of many problems in traditional networking with introduction of virtualization of resources [2] Internet Engineering Task Force (IETF) started working in decoupling data and control plane [3] that helps virtualization of resources, network management, and development of applications without changing underlying hardware SDN can be used as overly for IoT into the real world It can adapt the current system in support of different IoT challenges including routing, adaptability, heterogeneity, scalability, and network cost In the following sections, we will see the comparison of traditional network with dynamic Software Defined Network We will focus on the motivation behind using Software Defined Network for Internet of Things Subsequently, the study of emerging architecture of SDN with IoT Lastly the areas, where Software Defined Networking is used to overcome the challenges in IoT including security, intelligent routing, QoS provisioning, and Load Balancing 2 Literature Review IoT applications are providing quality in communication among human and machine Lack of flexibility, agility, and programmability in IoT bringing network scalability and heterogeneity challenges IETF first started working in decoupling data and control plane is presented in [3].it helps virtualization of resources, network management and development of applications without changing underlying hardware SDN provides centralized management of data through intelligent entity known as SDN controller The work in [4] presented multi SDN controller framework to handle network efficiently Multi SDN controller can not only handle diverse network with dedicated controllers but also monitor incoming and outgoing traffic efficiently On the other hand, the work in [6] shows that latency can also be reduced by installing multiple controllers SDN based security for IoT in [4] classified security-based solution into three ways including network based, traffic based, and crypto based solution Cryptography-based solutions presented in [7] which is concerned with confidentiality of data and ownership Symmetric (AES), Asymmetric algorithms (RSA) algorithms are used for encoding and decoding of data The work in [8],[9] envisioned the structural or topological properties of network Authenticity and integrity of information is merely depending on correct value of consensus In [10] author describe the delay minimization method for controlling IoT traffic through SDN Delay minimization is composed of three main functions 1) path resolving 2) delay tracking 3) delay management Integration of SDN into 6LowPAN purpose a new architecture SD-6LowPAN to obtain softwarized Network for multihop forwarding is described in [11] Where, SDN layers provide network programmability for applying forwarding rules through centralized SDN controller Software Defined Networking (SDN) In recent years, Software Defined Networking has made new paradigm in network management [3],[4] IoT provide opportunities to solve rigidity of traditional network In traditional networking data plane and control plane were combined in each device including routers, load balancer and switches as shown in Fig Data Plane is responsible for forwarding and processing of data packets IoT selects the next hope based on instructions given by control logic along the path Data plane forwards packet through network devices and decides what to do, where to move and whether to encapsulate or decapsulate data Control Plane is also called network signaling It provides the management of traffic using its control logic IoT also provides the network management interface for network configuration Control plane consists of routing table where the next hope routing information is stored and updated accordingly Some of traditional networking limitations are [4],[6]: • • They are tightly coupled (control plane can’t be used separately and vice versa) Very costly • • • New features can only be added IoT the vendors will, and time frame of adaptability is very large No centralized management and each device managed and configured individually Innovations are rare In the domain of Networking not so many innovations are found in recent time as compared to computing industry Figure 1: Traditional Networking [2] 3.1 SDN Architecture Software Defined Networking (SDN) come up with the solution of many problems in traditional networking with introduction of virtualization of resources [6] Internet Engineering Task Force (IETF) started working in decoupling data and control plane in 2004[wiki] Which was later refined by Stanford’s Computer Science department in 2008 and introduced standard application programming interface (API) for controlling data plane from control plane In 2009 a protocol named OpenFlow, NFV were introduced for SDN and afterwards many other protocols like Broader gateway protocol (BGP), NETCONF etc were introduced In 2017 and 2018 multi SDN controller framework is introduced to handle network efficiently [4][7] SDN allows network administrators to handle changes in network dynamically SDN provides centralized management of data through intelligent entity known as SDN controller [3] SDN decouples software with hardware This allow management, optimization, security and configuration of devices by using flexible application programing interfaces (API’ The controller can update, IoT or delete the entries in flow table dynamically SDN architecture is shown in Fig IoT is divided into five parts and each part detail is as follows: Application Plane is also known as management plane consists of different kind of networking applications like load balancing, routing, firewall and network management Also, responsible for defining protocols for network traffic Figure 2: SDN Architecture [6] Northbound interface provides an interface between control plane and application plane IoT receives instructions from application plane and forward to control plane for southbound interface Control plane consists of SDN controllers who have capability to provide overall view of network and provide centralized network management SDN controllers also provide high security mechanism by applying different cryptography algorithms, firewalls and device identification schemes This allocate resources dynamically and facilitate high data transmission for fulfilling network needs Southbound interface provides an interface between control plane and data plane IoT receives instructions from controllers for underlying hardware devices Also, gives communication protocol for controller and data plane known as OpenFlow protocol OpenFlow protocol mange the routing information and load balancing in data plane Data plane consists of infrastructure for physical and virtual network devices, used forwarding data packets from source to destination Network devices are interconnected through wired or wireless communication medium Table [6]: Software Defined Networking vs Traditional Networking Sr.no Criteria Network management Global view Maintenance cost Updating/Error handling time Adaptation/scalabilit y Recourse utilization Traditional Networking SDN Difficult (changes are made IoT each Easier device separately) Not possible Central view controller High Low High (sometimes take months) Low Not possible Highly possible Less High of Handling ability of Difficult Large/Heterogenous Networks Less Difficult Brief overview of Internet of Things (IoT) In this era of Globalization, internet of things has taken much attraction from both educational and commercial organization perspective This is because, we can manage objects remotely, efficiently and quickly This is capable to connect anything IoT anyplace and anytime with anyone by using wired or wireless path/network [1] IoT applications are everywhere including home automation, smart cities, livestock and logistics, transport management and smart grids This increased human quality of life because everything is managed on one click Internet of Things give the notion of developing infrastructure for connecting virtual and physical objects IoT evolved different technologies such as software defined network (SDN), Cloud Computing, Artificial intelligence and Wireless Sensor Networks (WSN) to get more benefits from such emergence [2] In this report, our focus is to see the benefits from emergence SDN with IoT This has been a hot area for researchers during last decade According to google scholar the number of journals and conference papers has been doubled during 2010 to 2017[3] 4.1 IoT Architecture IoT World forum committee release the standardized architecture in 2014 This model is used as a common architecture by most of the researchers as a framework IoT architecture is shown in Fig IoT consists of following parts: IoT Objects sense the data from the environment using sensors The sensed data is sent and receive among them using network devices such as routers, switches, firewall and load balancer that forward data to computing devices using IoT gateway IoT gateway is an interface between IoT devices and IoT cloud The purpose of this gateway is to convert incoming data flows into suitable information required for processing IoT cloud is a computational place where all the data is stored and processed Applications are run on electrical devices such as phone or PC that are responsible for interpretation of information IoT consists of different applications that use IoT input data or control IoT devices Figure 3: IoT Architecture[6] 4.2 Challenges of IoT There are many challenges that hinders the successful deployment of IoT application Some of the challenges are discussed as follows Big data: Internet of things connect billions of devices around the globe that create massive amount of data Reliability of applications is merely depending upon the management and integrity of data This is a challenge to dealt with such huge amount of data and heavily depend on effect data management handling Adaptability/Scalability: Number of devices in IoT are growing rapidly Current IoT environment is not enough to adapt changes to resolve scalability issues Scalability demands that environment should change according to demands and needs In the absence of scalability system become rigid and faces many challenges whenever some adaptation is required Heterogeneity: IoT connect thousands of devices and network with different architecture, operating systems and protocols they used IoT has become difficult to handle such heterogenous environment as the number of devices are growing rapidly Privacy & Security: IoT applications success is strongly depend on the privacy and protection of data from unauthorized persons Different security mechanism was given but a reliable security structure still lacks due to constrained network devices in term of energy, computational power and storage size The global acceptance of IoT and its related technologies depend on security and privacy of user data Because of heterogeneity and complex structure IoT still lacks in security So, a powerful, reliable and effective security mechanism is still needed in IoT Interpretability: IoT environment connect almost every object through a common network framework As number of IoT devices are increasing the lack of interpretability between them has become a critical issue Although many suggested solutions resolved some issues such as data over sound technology but still interpretability is a big issue in IoT system 5 SDN with IoT SDN and IoT are two distinct technologies, where SDN provides the flexibility in traditional networking by decoupling Data plane (hardware) and Control plane(software) and communication is done through application programming interfaces (APIs) like OpenFlow [2a8] SDN helps virtualization of resources, network management and development of applications without changing underlying hardware The focus is on application side with centralized management of network (SDN controller) On the other side, aim of IoT is to connect billions of objects with internet This is difficult to manage huge, heterogenous network of IoT SDN controller can not only handle such kind of diverse network but also monitor incoming and outgoing traffic[3] Also, confirm security of network from inside and outside attack by applying advanced security mechanisms SDN can adapt the current system in support of different IoT challenges including routing, adaptability, heterogeneity, scalability and network cost 5.1 SDN based IoT Architecture SDN can be emerge with IoT as an overly to get benefits in the real world The fig shows the SDN based IoT architecture IoT framework is interconnected with SDN controller who manage the underlying heterogenous network Fig consists of the following components IoT object consists of IoT agents who are responsible for sensing, collecting and forwarding data to IoT controllers Each IoT object must be registered with IoT controller before start working through their Unique Addresses, object identifier no and share information like used network communication protocol Figure 4: SDN based IoT Architecture [7] IoT controller registers IoT agents with their addresses and on receiving data forwarding request IoT develops forwarding rule depending upon communication protocol used and share such rules with SDN controller This is responsible for finding destination object address and take necessary actions on data received from the IoT agents SDN controller is responsible for establishing connection between both objects and decide path by running different routing algorithms It also manages incoming and outgoing traffic in SDNenabled network SDN controller can efficiently secure the IoT network by implementing security firewalls, cryptographical algorithms, user authenticating mechanisms despite of their latency overhead In recent research [3][7], it is observed that such latency can be reduced by installing multiple SDN controllers to increase network performance 5.2 Security for SDN based IoT IoT contains constrained devices like sensing agents the make IoT more vulnerable to security risks Due to limited energy, computational power and storage size implementing security mechanisms such as cryptography algorithms, authentication procedure brings down the network performance SDN based security for IoT in [4] classified security-based solution into three ways including network based, traffic based and crypto based solution cryptography-based solutions are concerned with confidentiality of data and ownership Symmetric (AES), Asymmetric algorithms (RSA) used for encoding and decoding of data It helps to mitigate shared identity by trusted third party certificate authority which encrypts not only meta data but also source and destination IP addresses The drawback is communication overhead because of the control traffic between SDN controller and IoT agents On the other side, Network based solution dealt with the structural or topological properties of network Authenticity and integrity of information is merely depending on correct value of consensus Resilient formation of nodes in network is necessary to achieve robustness Lastly, focus in traffic-based solution is on flow of data among nodes in distributed system These security mechanisms confirm the flow in the presence of malicious activities This is achieved by running different algorithms like OpenFlow Table 2: SDN based IoT security solutions Sr.no Search effort Black SDN for the Internet of Things[7] achievement Mitigate against data gathering attacks Encrypting payload including source and destination IP addresses Flow based security for IoT devices using SDN Guarantee the independence gateway [8] of each domain in case of failure SDN based IoT architecture for security Propose routing protocol for improvement[9] distributed environment Synchronization among multiple controller to achieve combined objective 5.3 Dynamic Load Balancing (LB) for IoT Load Balancing provides data distribution services among different resources of network in order to maximize the optimization of resources Load balancing methods can be static or dynamic In static methods the LB is implemented on hardware level and the behavior of the user cannot be predicted On the other side, dynamic LB is more efficient because the load is distributed based on different policies and protocols A suitable load balancing helps in maximum scalability, minimum response time and maximize throughput Centralized network management and global view from SDN controller allows allocation of resources through Southbound APIs Network administrator can implement rules and policies for LB at application layer For load balancing two operational rules are defined 1) Reactive 2) proactive In former approach data packets are handled by network devices and forwarded without involvement of centralized management On the other side, in later approach devices send data packets to centralized controller for decision making In such a way load on SDN switches can be adjusted dynamically due to global view of network SDN controller collects data from IoT devices and decide the routes based on different protocol algorithms [8]such as Round Robin (RR), Weighted Round Robin (WRR), Least Connection (LC), Weighted Least Connection (WLC), Contention Balancing (CB) etc with RR algorithm a round like table is managed which hold the node information of network and each node gets the task equally likely but it does not bring the characterization of node into consideration While, WRR algorithm assigns weight to each node calculated based on node specification e.g processing power, number of connections, battery specification etc Least connection assigns a value to node based on number of connections The table is updated whenever a connection is added or deleted The WLC use weights with LC and assign more connection to that node having more weight value Lastly, contention Balancing algorithm checks the usage of network devices and when it reaches to 80% of the total capacity then load balancer module is activated to overcome this traffic bottleneck and controller finds the new best path and route the upcoming traffic to this new path Table 3: Dynamic Load Balancing algorithm implemented in SDN based IOT Sr.no LB Algorithm RR Working A round like table is managed which hold the node information of network and each node gets the task equally likely WRR Assigns weight to each node calculated based on node specification e.g processing power, number of connections, battery specification etc LC WLC CD Assign value to node based on number of connections The table entry is updated whenever a connection is added or deleted It uses weights with LC and assign more connection to that node having more weight value checks the usage of network devices and when it reaches to 80% of the total capacity 5.4 Minimizing latency of critical traffic through SDN Huge amount of traffic is generating by billions of devices in IoT This introduce network management challenge for operators In such situation, keeping low end to end flow latency is crucial SDN can manage this critical end to end IoT traffic efficiently For the sake of handling latency, one of the important parameters is the identification of data flow paths and end to end delay along identified paths By identification of highly delayed paths traffic can be redirected through the paths with minimum delay In [minim latency] describe the delay minimization method for controlling IoT traffic through SDN Delay minimization is composed of three main functions 1) path resolving 2) delay tracking 3) delay management Delay tracking and path resolving methods run in parallel to monitor activities while delay management runs on top of them and take necessary actions based on controlled information Path resolving creates a list of a variable paths between the heads and tail node To develop list, it picks up sending and receiving IoT host and the head and tail nodes of available paths Afterwards, it creates links sequentially between each consecutive node This is done by brute force algorithm Delay tracking measures the real time latency of identified network paths This provide the estimation of delay that a data flow will face between the head and the tail nodes of sending and receiving hosts Based on estimation, the route between two hosts is decided All the implementation functions run in SDN controller A request is firstly sent for probe to SDN controller who selects route based on the decision taken by delay management function Delay information along traveled path is updated in the list after completion of successful communication Delay management runs on top of both paths resolving and delay tracking It applies a function that gather takes necessary actions based on controlled information and select real time decisions to select the most appropriate path among all available to get the latency minimum Delay minimization function consists of following steps that are executed iteratively: 1) Gather list of paths for pairs of IP addresses from path resolving function 2) Gather recent value of delay from delay tracking function 3) Find best suitable path based on min= (d1,d2,d3……dn) 4) Program path in head and tail node direction 5) Update data flow path action list 5.5 Multi-hop Forwarding in 6LowPAN through SDN The development of Wireless sensors network (WSN) has been immensely increased in recent years WSN plays an important role in the expansion of Internet of Things market for low power and low-cost networks Typically, traffic patterns were point to multipoint i-e sink to sensor nodes and multipoint to point i-e sensor nodes to sink Nowadays, application in industrial system has introduced new traffic pattern machine to machine (M2M) 6LowPAN has been developed to enable end to end communication in low power and lossy network Lack of standardization layer protocols arise problem in nodes to build formalize table There is need to establish an adohoc system for network configuration to meet users need On the other hand, Software Defined Networking (SDN) is capable to manage configuration and interpretability through centralized manner In recent years, SDN has applied softwarization approach in wireless sensor ntwork domain.Integration of SDN into 6LowPAN purpose a new architecture SD-6LowPAN to obtain softwarized Network for multi-hop forwarding SD-6LowPAN architecture is described in fig It consists of three parts 1) SDN controller 2) Gateway 3) SDN Node, SDN Controller gets the information related to 6LowPAN during formation of the network and complete flow tables at each sensor node through SBI interface Every packet is forwarded according to mesh under technique It manages 6LowPAN operations with the help of Local Controller SDN Controller gather information regarding network formation through the gateway to generate the global view of network Afterwards, provide the forwarding policies according to configuration rules through SBI interface to SDN Node Gateway allows the interaction and exchange of information between SDN Controller and SDN Node All the information regarding construction DoDAG formation and maintenance is exchanged through gateway SDN Node is constrained sensor node of 6LowPAN with the additional SDN sub-layer that intercept the packet flow between 6LowPAN and MAC layers SDN layers also provide network programmability for applying forwarding rules through centralized SDN controller All the operations are handled by flow tables at SDN Node SD-6LowPAN architecture allows dynamic and control multi hop forwarding over 6LowPAN through software defined networking in 6LowPAN It presents a new forwarding paradigm in Figure 5[9]: System Architecture SD-6LoWPAN networking to achieve network reconfigurability SDN inclusion provide low cost, better performance in communication between end point Conclusion IoT is changing the way of communication among different physical objects with the internet IoT applications are providing quality in communication among human and machine Lack of flexibility, agility and programmability in IoT bringing network scalability and heterogeneity challenges It was needed to use SDN technology for IoT management In this report, we presented the need of SDN for providing solution of many problems in traditional networking with introduction of softwarization and virtualization of resources In this work, we studied the motivation behind emerging these two different technologies Lastly, we make a comparative study from different perspective to see how SDN can adapt the current system in support of tackling different IoT challenges References [1] H.Atlam, J.Robert, B.Wills “Internet of Things: State-of-the-Art, Challenges, Applications, and Open Issues.” International Journal of Intelligent Computing Research volume 9, no pp 1-12 (March 2018) [2] M.Shailendra, A.Rahman “Software Defined Networking: Research Issues, Challenges and Opportunities.” Indian Journal of Science and Technology volume 10, no 29 pp 1-9 (February 1, 2017) [3] Tayyaba.S.Khan, Munam.A.Shah, Omair Khan, Abdul.W.Ahmed “Software Defined Network (SDN) Based Internet of Things (IoT): A Road Ahead.” In Proceedings of the International Conference on Future Networks and Distributed Systems - ICFNDS vol 17, pp 1–8 Cambridge, United Kingdom: ACM Press, (2017) [4] K.Kubra, S.Zeadally “Securing Internet of Things with Software Defined Networking.” IEEE Communications Magazine volume 56, no pp 186–192 (September 2018) [5] Singh.A.Kumar,S Srivastava “A Survey and Classification of Controller Placement Problem in SDN: A Survey and Classification of Controller Placement Problem in SDN.” International Journal of Network Management Volume 28, no pp 512-521 (May 2018): [6] Vandana C.P “Security improvement in IoT based on Software Defined Networking (SDN).” International Journal of Science, Engineering and Technology Research (IJSETR), Volume 5, no 1, (January 2016) [7] S.Chakrabarty, D W Engels, and S Thathapudi “Black SDN for the Internet of Things IEEE 12th International Conference on Mobile Ad Hoc and Sensor Systems, pp 190–98 (2015) [8] B.Peter, R Austin, E Popov, M Sharma, and R Watson “Flow Based Security for IoT Devices Using an SDN Gateway.IEEE 4th International Conference on Future Internet of Things and Cloud (FiCloud), pp 157–63 (2016) [9] G.Tanganeli, V.Antonio, E.Mingozzi “Enabling Multi-Hop Forwarding in 6LoWPANs through Software-Defined Networking,” pp 211-222 (2019)