Wireless sensor networks protocols in IoT. A performance evaluation and comparison

Table of Contents

ABSTRACT

Table of Contents

List of Figures

List of Tables

CHAPTER 1 - INTRODUCTION
1.1 Problem statement
1.2 Aim of study
1.3 Research questions or hypothesis
1.4 Significance of study
1.5 Thesis outline

CHAPTER 2 - LITERATURE REVIEW
2.1 Internet of Things
2.2 Applications of IoT
2.2.1 Smart Homes
2.2.2 Smart City
2.2.3 Smart Grid
2.2.4 Industrial Automation
2.2.5 Wearables
2.2.6 Smart Agriculture
2.2.7 A Short Study on the new CIU Science & Technology Building
2.3 Issues in IoT
2.4 Routing in IoT
2.4.1 Data Routing Issues
2.4.2 Routing Protocols and Techniques in IoT
2.5 Why AODV, DSR and OLSR?
2.6 Simulation

CHAPTER 3 - RESEARCH METHODOLOGY
3.1 Simulation of AODV, DSR and OLSR
3.2 OPNET Modeler Suite
3.2.2 Hierarchical Structure of OPNET
3.2.3 MANET Node Models in OPNET
3.2.4 Routing Protocol Configurations
3.3 Simulation of AODV using an IoT Scenario
3.4 Simulation of OLSR using an IoT Scenario
3.5 Simulation of DSR using an IoT Scenario
3.6 Network Simulator - 3 (NS3)
3.7 Simulation of AODV, DSR and OLSR using NS3

CHAPTER 4 - DATA ANALYSIS AND DISCUSSION OF FINDINGS
4.1 Performance Metrics
4.1.1 Routing Overhead
4.1.2 Average End to End Delay
4.1.3 Throughput
4.1.4 Mobility and Distribution of Nodes
4.1.5 Parameters for OPNET Simulation
4.1.6 Simulation Parameters for NS3 Simulation
4.2 OPNET Simulation Results
4.3 Discussion of OPNET Results
4.3.1 Discussion on Routing Overhead
4.3.2 Discussion on Average End to End Delay
4.3.3 Discussion on Throughput
4.3.4 Discussion on the Routing Protocols
4.4 NS3 Simulation Results
4.5 Discussion of NS3 Results

CHAPTER 5 - CONCLUSION
5.1 Conclusion
5.2 Recommendation
5.3 Future Work

REFERENCES

ABSTRACT

IoT has continue to grow bigger since from its inception. Many mobile devices are now available, the internet and its application have only grown bigger and better. As IoT is continually growing, so also is the complexity, as a result issues pertaining routing have also increased. Many researches have been made in attempt to proffer solutions that will either minimize or eliminate these routing issues. Different routing protocols have been designed with different specifications for different applications of the IoT. Also, attempts have been made to implement routing protocols of other types of networks in the IoT.

In this thesis, three Wireless Sensor Networks - Ad-hoc On-Demand Distance Vector, Dynamic Source routing protocol and Optimized Link State routing protocol have been simulated and compared in typical IoT scenarios. Their performance was evaluated using three performance metrics and then they were compared; the performance metrics are Routing Overhead, Average End to End Delay and Throughput. Different number of nodes with different percentages of mobile nodes were analyzed. Specifically, number of nodes analyzed were 20, 40, 60 and 70 with the number of mobile nodes 10, 15 and 20 using OPNET while with NS 3 20, 60 and 100 nodes were analyzed. For each of the number of nodes, all the number of mobile nodes were evaluated. The routing protocols were analyzed using the OPNET Simulation Software and NS-3and the environment size for the simulation was 1000m by 1000m.

Keywords: IoT, WSN, routing protocols, AODV, DSR, OLSR, OPNET, NS-3

List of Figures

Figure 1.1 Routing of data through nodes all connected

Figure 2.1 Some applications of the IoT

Figure 2.2 Different devices & sensors in a smart home 59

Figure 2.3 Different components of the smart city 60

Figure 2.4 Smart city as a small part of the future smart country 61

Figure 2.5 The smart grid 62

Figure 2.6 Structure of the smart grid 63

Figure 2.7 Automated machines in a manufacturing plant 64

Figure 2.9 A sportsperson using wearables to get information 65

Figure 2.10: A DAG, root node and leaf nodes

Figure 2.11: A DODAG with all leaf nodes directed towards S (root)

Figure 3.1 OPNET Academic version

Figure 3.2 General Workflow of simulations in OPNET

Figure 3.3 Project cycle in OPNET

Figure 3.4 Object Palette in OPNET

Figure 3.5 Routing protocol configuration for OLSR

Figure 4.1: AODV with 20 mobile nodes out of 60

Figure 4.2: DSR with 20 mobile nodes out of 60

Figure 4.3: OLSR with 20 mobile nodes out of 60

Figure 4.4: OLSR with 15 mobile nodes out of 60

Figure 4.5: DSR with 15 mobile nodes out of 60

Figure 4.6: AODV with 15 mobile nodes out of 60

Figure 4.7: AODV with 10 mobile nodes out of 60

Figure 4.8: OLSR with 10 mobile nodes out of 60

Figure 4.9: DSR with 10 mobile nodes out of 60

Figure 4.10: AODV with 20 mobile nodes out of 20

Figure 4.11: OLSR with 20 mobile nodes out of 20

Figure 4.12: DSR with 20 mobile nodes out of 20

Figure 4.13: AODV with 15 mobile nodes out of 20

Figure 4.14: DSR with 15 mobile nodes out of 20

Figure 4.15: OLSR with 15 mobile nodes out of 20

Figure 4.16: OLSR with 10 mobile nodes out of 20

Figure 4.17: DSR with 10 mobile nodes out of 20

Figure 4.18: AODV with 10 mobile nodes out of 20

Figure 4.19: AODV with 10 mobile nodes out of 40

Figure 4.20: DSR with 10 mobile nodes out of 40

Figure 4.21: OLSR with 10 mobile nodes out of 40

Figure 4.22: OLSR with 15 mobile nodes out of 40

Figure 4.23: DSR with 15 mobile nodes out of 40

Figure 4.24: AODV with 15 mobile nodes out of 40

Figure 4.25: AODV with 20 mobile nodes out of 40

Figure 4.26: DSR with 20 mobile nodes out of 40

Figure 4.27: OLSR with 20 mobile nodes out of 40

Figure 4.28: OLSR with 20 mobile nodes out of 70

Figure 4.29: DSR with 20 mobile nodes out of 70

Figure 4.30: AODV with 20 mobile nodes out of 70

Figure 4.31: AODV with 15 mobile nodes out of 70

Figure 4.32: DSR with 15 mobile nodes out of 70

Figure 4.33: OLSR with 15 mobile nodes out of 70

Figure 4.34: OLSR with 10 mobile nodes out of 70

Figure 4.35: DSR with 10 mobile nodes out of 70

Figure 4.36: AODV with 10 mobile nodes out of 70

Figure 4.28 Routing overhead for number of mobile nodes

Figure 4.29 Routing overhead for number of nodes

Figure 4.30: Average End to End Delay for number of mobile nodes

Figure 4.31: Average End to End Delay for number of nodes

Figure 4.32: Throughput for varying number of mobile nodes

Figure 4.33: Throughput for number of nodes

Figure 4.34 Throughput for 20 nodes - Using NS3

Figure 4.35 Throughput for 60 nodes - Using NS3

Figure 4.36 Throughput for 100 nodes - Using NS3

Figure 4.37 NS3 Simulation Results for Average Throughput

List of Tables

Table 4.1: Routing overheads for 20 nodes

Table 4.2: Routing overheads for 40 nodes

Table 4.3: Routing overheads for 60 nodes

Table 4.4: Routing overheads for 70 nodes

Table 4.5: Average End to End Delay for 20 nodes

Table 4.6: Average End to End Delay for 40 nodes

Table 4.7: Average End to End Delay for 60 nodes

Table 4.8: Average End to End Delay for 70 nodes

Table 4.9: Throughput for 20 nodes

Table 4.10: Throughput for 40 nodes

Table 4.11: Throughput for 60 nodes

Table 4.12: Throughput for 70 nodes

Table 4.13: Routing overhead for number of mobile nodes

Table 4.14: Routing overhead for number of nodes

Table 4.15: Average End to End Delay for number of mobile nodes

Table 4.16: Average End to End Delay for number of nodes

Table 4.17: Throughput for number of mobile nodes

Table 4.18: Throughput for number of nodes

Table 4.19 Average Throughput from NS3

CHAPTER 1 - INTRODUCTION

Internet of Things (IoT) has grown over the years in both size and complexity. These complexities have continually grown because of homogeneity of devices and network standards. This has brought along with it so many issues that researchers have been working on continually with a specific end goal to make IoT much more better. Some of these issues and problems are scalability, mobility, security, routing of data, privacy, e.t.c.

Routing of data is an extremely big issue in IoT; this is because data itself is is a standout amongst the most critical things in any system. Routing of data involves taking data from an end device to another end device by utilizing the correct and most efficient routes available.

In IoT the data being sent being sent between different devices is very important; also important is the integrity of the data, security, availability and scalability. All these are important but are big issues in the routing of data because the devices are mostly heterogeneous and so are the networks. Due to the differences in the device types, networks, memory size and power consumption, it is difficult (and becoming even more difficult) to route data efficiently and efficaciously to transmit data from one device to another. In order to make IoT in general better, the problems involved in data must be solved and in addition the other issues.

1.1 Problem statement

As the IoT has continually grown bigger, so also has the data routing problems increased greatly. Some of these issues are security, scalability, mobility, availability, context awareness and lots greater. Many protocols have been created over the years and have also been improved multiple times. With all these issues nodes (IoT devices) connected (directly or indirectly) as shown in Figure 1.1, there is a need for more research in order to make more improvements or even make better and more effective protocols to make routing of data in IoT better.

Abbildung in dieser Leseprobe nicht enthalten

Figure 1.1 Routing of data through nodes all connected

1.2 Aim of study

In this thesis, the main aim is to analyze three routing protocols used in wireless sensor networks - Ad-hoc On-Demand Distance Vector, Dynamic Source Routing protocol and Optimized Link State Routing protocol, implement them in IoT scenarios and then see if they can be used for IoT and the circumstances for which they can be used.

1.3 Research questions or hypothesis

Basically, the main question that will be asked in this research is can these Wireless Sensor Networks (WSN) routing protocols effectively be used in IoT? From the main question others spring up, one of which is: if they can be used, under which circumstances can they be used?

1.4 Significance of study

This research will provide more information about WSN routing protocols and how they can be used in IoT; in other words, it will provide more routing solutions or alternatives to some of the IoT routing protocols already in use.

1.5 Thesis outline

This section will provide an outline of the thesis, each chapter and the subject of discussion. The other chapters in this thesis are summarised in the preceding paragraphs.

Chapter Two: Literature review. This chapter will discuss different papers and journals related to IoT, routing of data in IoT, issues of the routing of data in IoT and protocols used. Also, other works previously done will be thoroughly discussed in order to lay a proper foundation for this work.

Chapter Three: Research methodology. In this chapter, the different simulations that were done on the protocols will be discussed. Basically, the simulators - Optimized Network Engineering Tools (OPNET) Modeler and NS-3will also be discussed in depth.

Chapter Four: Data analysis and discussion of data findings. In this chapter all the results from the simulations done in OPNET and NS-3 will be analysed and discussed, inferences will then be made from the findings. These inferences will then be used to decide on improvements that will or can be made so that data routing in IoT can be better.

Chapter Five: Conclusion. This chapter concludes the thesis, also future works and recommendations will be discussed here.

CHAPTER 2 - LITERATURE REVIEW

2.1 Internet of Things

In this chapter Internet of things (IoT) will be discussed in detail. Additionally, the issues of different components of IoT will be discussed. Reviews will also be done on various papers and journals to expose what data routing in IoT really is and the problems it is facing; already preferred solutions and protocols will be reviewed.

IoT in the simplest terms is an interconnection of different things which have been made to be smart and do communicate with one another without the help (or with a little) of a human. From the inception of the IoT, there have been many definitions by different people all in an attempt to correctly describe what the IoT really is. A commonly accepted definition for IoT is given as: “a dynamic global network infrastructure with self-configuring capabilities based on standard and interoperable communication protocols where physical and virtual things have identities, physical attributes and virtual personalities, and use intelligent interfaces and are seamlessly integrated into the communication network” 11.

2.2 Applications of IoT

There have been different classifications of the applications of IoT according to different people based on some facts or reasons. Some of these applications will be discussed in the next sections. Currently, there are many aspects of applications of the IoT buy only a few of them are being explored presently 8. Some of these applications are shown in Figure 2.1. In this section, some of these applications will be discussed.

Abbildung in dieser Leseprobe nicht enthalten

Figure 2.1 Some applications of the IoT

2.2.1 Smart Homes

The most popular utility of the IoT is the smart home - also called home automation; in this thesis, smart home and home automation will be used interchangeably, they have the same meaning; which is, the application of the internet of things within the home environment 6. According to 1 home automation as the use of the internet in the home environment so that the occupants can enjoy more convenience, comfort, energy efficiency and security.

Basically, the idea of the smart home or home automation is to make the ordinary home appliances and equipment such as refrigerators, beds, windows, air conditioners, etc. to become smart, then communicate with each other in order to do a particular service for the occupant of the home. Just like the cell is the simplest unit of life, the smart home is the basic unit of the smart city.

By incorporating advanced IoT technologies into buildings, the consumption rate of resources such as water and electricity will be reduced greatly; also the level of satisfaction of the people living in these buildings will also be improved 40. Figure 2.2 shows different sensors, smart devices and the service they provide in a smart home.

Abbildung in dieser Leseprobe nicht enthalten

Figure 2.2 Different devices & sensors in a smart home 59

2.2.2 Smart City

According to 40, Smart city is an environment which is cyber-physical (In line with Wikipedia, cyber-physical is a mechanism controlled by computer-based algorithms which are tightly integrated with the internet and its users) and is developed by using advanced communication facilities and services over massive urban regions. The smart city application of the IoT is more than just managing flows of the urban areas but also to allow real time (instant or very quick) response to challenges faced by the occupants of those cities 8. Figure 2.3 shows a generally of a typical smart city in terms of its components and services provided.

Abbildung in dieser Leseprobe nicht enthalten

Figure 2.3 Different components of the smart city [60]

According to 12, the smart city has some smart activities such as smart monitoring of the quality of air, watering of gardens, smart lighting of the city, discovering emergency routes for special cases like accidents and fire outbreak, and many more smart features. In a smart city, the aim is to make sure that all components of the city are to some certain degree smart so that the benefits can be enjoyed by the citizens of that city. The smart city components consist of all infrastructure such as roads, smart homes, filling stations, traffic lights, head lights, waste or dust bins, and much more.

“A Smart City is one that places humans at the centre of development, incorporates Information and Communication Technologies (ICTs) into urban management, and makes use of these elements as tools to stimulate collaborative planning and citizen participation. By promoting integrated and sustainable development, Smart Cities become more innovative, competitive, attractive, and resilient, thus improving lives“ 41.

The smart city according to CISCO 2 is just a small part of a big internet of things project which has been projected to surface in the nearest future. This is shown in Figure 2.4, it can be seen that in the near future not only homes or cities are expected to be smart, but a whole country and then the whole world.

Abbildung in dieser Leseprobe nicht enthalten

Figure 2.4 Smart city as a small part of the future smart country [61]

2.2.3 Smart Grid

The smart grid is a building block for management of energy which is a significant prerequisite for the sustainability of an environment 9. According to 54, the smart grid is an intelligent electrical distributive system that delivers energy flows from the producers to the consumers in a bidirectional way. In addition to being intelligent, the smart grid is also a digitised network for delivering electricity in an optimal way from the producer or source to the consumer 3. The smart grid is shown in Figure 2.5. It is characterised by a two-way flow of electricity and information with the capability of monitoring and responding to changes that occur in power plants, consumer preferences and individual appliances.

Abbildung in dieser Leseprobe nicht enthalten

Figure 2.5 The smart grid [62]

The smart grid has self-healing capabilities to correct problems or issues that arise during operations, this makes sure that it is always reliable and immune to unexpected failures 3. Also, the smart grid works in a similar fashion with the internet. It allows multiple users and applications at the same time and they can operate together without problems; the smart grid also allows these couple of users to operate from any distances using as much infrastructure as possible without possibly noticing the distance.

Abbildung in dieser Leseprobe nicht enthalten

Figure 2.6 Structure of the smart grid [63]

The smart grid does what the traditional grids do but in a more effective, efficient and smarter way. In the traditional grids, power is generated by a few central power plants and then disseminated to the consumers using substations, transformers and cables but the smart does not, in fact, the producers can also be final consumers (just like a peer-to-peer system). The consumers can also send generated power from their micro grids (using media like solar panels, wind turbines) to the smart grid; the grid then manages the power using some smart energy control services and stores it in designated energy stores 9. Basically, the goal of a smart grid is to optimise efficiency, monitor problems to reduce outages, integrate renewable power generation and reduce power peak demands 3. Figure 2.6 shows a simple structure of the smart grid.

The smart grid has some additional applications such as monitoring and exchanging of information on energy flows; with this, it minimises losses and increases efficiency thereby disseminating the needed amount of power generated by the consumers when need be. This is achieved by using advanced metering infrastructure which includes smart meters and automatic control devices, smart watches, smart appliances and variable prices of electricity. With the devices mentioned the smart grid can get all related information and then use it in advance to determine the expected demand of consumers and to adjust the consumption and production of electricity accordingly. This helps to avoid overloads, possible blackouts and instant action in any case of failure.

The information generated by the smart grid is usually sent to the consumers, this will increase their awareness about consumption of energy and also motivate them to always check and manage their habitswhen it comes to consumption.

2.2.4 Industrial Automation

Industrial automation is an application of the IoT which involves making machines used for manufacturing in industries to talk to one another in order to speed up or to improve the quality production. These machines are made to be smart so that they can optimise data so as to improve their performance. Figure 2.7 shows a set of automated machines lined in a manufacturing plant. Industrial automation is a vast connection of advanced sensing, computing, communication and actuation. Also, the things in the industrial automation are called network nodes and trust worthy time information is very critical. This means that accurate time stamps of data need to be taken at all times if this is not done serious damage could occur which will be very detrimental to the companies.

Industrial automation is used in lots of industries for various purposes. A good example is in the automotive industry for manufacturing activities such as managing a fleet of cars. With IoT, these organisations can monitor the cars, their environmental performance and also use the data gathered to decide which ones need repairs or maintenance 12. Industrial automation can be used in virtually all industrial activities, be it commercial or financial transactions, logistics, banking, monitoring of processes and much more 9.

Abbildung in dieser Leseprobe nicht enthalten

Figure 2.7 Automated machines in a manufacturing plant [64]

The IoT industrial automation is also used in the healthcare service industry 13. With IoT, there have been many inventions in health care sector which have improved greatly the way patients are being tended to 55. With the vast amount of sensors, communication and identification techniques that IoT provides, it is now possible to monitor all things (equipment, medicines, etc.) and people; this has made it possible for more effective diagnosis medical administration. It is now possible for all relevant information to be taken, tracked and monitored, and shared in order to provide better healthcare services. A good example is the use of sensors to collect a heart rate from a patient and sent to the doctor's office, the doctor can then use the information collected for better examination. The collection and tracking of this information are made possible by using personal or handheld computing devices such as phones and tablets which are constantly connected to the internet.

In the mining industry, IoT is also used to make mining safer for the miners 13. The safety of miners operating in underground mines has continually been a significant concern in the industry. In order to prevent or reduce accidents during mining, IoT is used in many ways to sense or forecast disasters, detect faults and cracks in order to make early warning signals for the miners when need be. Also with IoT effective communication between miners underground and the surface is improved so that the companies can track miners' location. Since the mines are usually exposed to diseases and other biological hazards, IoT sensors are used to acquire information which is used for early detection and diagnosis of these diseases.

Other applications of IoT in industrial automation exist. A few could employ the usage of IoT in transportation and logistics where vehicles are equipped with RFID tags, bar codes and sensors so that they may be tracked by the companies. Also in the production of cars, machines are automated to take charge of the whole manufacturing process.

2.2.5 Wearables

The advancement in IoT has made it possible for the integration of sensing and recognition abilities in smart wearables today 16. Wearables are devices that are incorporated into clothes, jewellery, shoes and even the human body with the aim of collecting and/or tracking data from the person wearing them 20. These wearables have components like sensors and modules embedded in them coupled with advanced programs; they are also built to consume as little power as possible, this enables consumers to carry them along and use them for some considerate amount of time. Figure 2.15 shows wearables and functions or services they provide for the person using them.

Wearables are usually small and with very little mass making them comfortable or easy for users to carry them around. The use of wearables is getting more and more popular every day, this because of the growth in computer and mobile technology. Mobile gadgets consisting of phones and tablets have end up more and more utilized by almost everyone 20. Another reason for the increased popularity of wearables is the increase in production of sensors and processing elements for cheap prices - this has helped to boost the chances of people wanting to use more mobile devices 20. Wearables are fast becoming integrals components of the health sector.

Another sector that is employing the use of wearables is sports. As seen in Figure 2.9 a sportsperson can have one or more wearable(s) on/in their body while doing whatever sports they participate in. This is done to collect data by different experts which in turn will be used to improve training or treatments or other research purposes related to the field or sport. The military is also another sector currently employing the use of wearables 20. Wearables are integrated with the soldiers' clothes and gear so that they can be monitored remotely.

Abbildung in dieser Leseprobe nicht enthalten

Figure 2.9 A sportsperson using wearables to get information [65]

2.2.6 Smart Agriculture

Some factors affecting the quality of products can be controlled by humans 27. Some of these factors include soil PH, moisture, humidity, temperature, e.t.c.; these factors can be controlled to improve the product quality by using IoT. Also, during harvesting period IoT can be used for automated grading and package, using proper identification and tracking methods. IoT can also be used to observe the state of the transport and storage facilities of the harvested products, thereby increasing or preserving the quality of these products.

Another important contribution of IoT in Agriculture is analytics for the farmers. Through the use of IoT, the farmers can gather information and discern which crop and time of the year to farm them in order to get the optimal results; in other words, decision making of farmers is improved greatly by monitoring everything about the farm and products using IoT 27 28.

As the study in India according to 29 shows, IoT can be used to improve security in farms. Security issues in farms for the crops could be in the form of pests, harsh weather, rodents, insects, e.t.c. To combat these, monitoring systems which include CCTV, sensors, trackers and much more are being installed in the farms.

2.2.7 A Short Study on the new CIU Science & Technology Building

With the whole world evidently transitioning into the use of the IoT and its applications to do different things, Cyprus International University (CIU) is not left out. The university has made top notch upgrades and growth over the last few years; this sort of improvements is the constructing of the new technology center (a smart building) inside the school. The constructing is a one-of-its-kind, it's miles the first of its kind in now not only Cyprus but the whole region as well. In this section, a brief review is made on the building, exposing its strengths and facilities and applications of the IoT.

The first essential characteristic of the building is the internet connectivity. It is equipped with efficient and effective internet so that all things can be able to communicate to other things either close by or far away. The internet is supplied and managed by the computer centre of the school. Every part of this magnificent building has fast internet connectivity - which is a major requirement of an efficient and effective internet of things.

The next important feature of this building is the connection between the things. Different things are connected in the building in order to provide a service or satisfy some particular needs of the users of the building; these connections are essentially in distinctive forms. Some connections are permanent (always connected to each other) while some are based on some triggers which have been set and defined by the engineers. Actually, the specifics of the triggers and other details are not given here, this is because all the information pertaining cannot be given out to just anybody. A terrific instance of associations between the things and a trigger circumstance is the fire alarm system. This system is made of sensors, these sensors are always sensing the environment; when there is a fire outbreak in any part of the building, the system will be triggered and a fire alarm will be sounded and the proper personnel involved will be alerted.

Right from its inception, users of IoT (both individuals and companies or organisations) and researchers have always been concerned about the security and integrity of data. The engineers responsible for the CIU technology centre has done some good work to ensure security is put in place. They have put monitoring systems in the building to track every activity that goes on in the activity. These monitoring systems are both physical and electronic/software, once more not much information is disclosed here but the systems have been put in place.

The other issue in IoT which researchers and users alike have always been concerned with is power and energy optimisation. This problem has almost totally been taken care of. The building is self-powered using both solar-generated electricity and the traditional power grid generated electricity. This will ensure that the building always has power, this means constant internet and connectivity which is good for effective application of the IoT.

Basically, the building has many sensors for different uses, is connected to the information centre for constant monitoring. When 100 percent completed and being put into use, it will be a state of the art smart building which will speed up research and development for the university, other universities and organisations in the region.

2.3 Issues in IoT

As was mentioned earlier, the IoT right from its inception has had to deal or cope with some issues. These issues have only increased greatly as the field has grown bigger over the years. There has been much work done by different researchers as they have tried in one way or the other to solve or reduce the impact of these problems. Different researchers have come up with different ideas on classifications of the issues IoT. Additionally, there are numerous open studies issues in the IoT which many researchers were running in recent times. These open research issues are normally the premise for further study on the primary problems of the IoT. According to 55, the open research issues in IoT can be classified as standards, mobility support, naming, transport protocol, traffic characterization & Quality of Service (QoS) support, authentication, data integrity, privacy and digital forgetting. These open research issues are not necessarily referring to the problems that are encountered in the applications of IoT, but they form the basis for further study on those problems. Basically, the major IoT problems or challenges are

- Mobility
- Reliability
- Scalability
- Management
- Availability
- Interoperability

2.4 Routing in IoT

In this section, issues or challenges which can be faced at some stage in routing in IoT will be discussed. Also to be discussed are some routing protocols that have been implemented for the IoT, including WSN routing protocols which can be used in IoT in some certain scenarios. Section 2.4.1 will discuss the routing issues in IoT and section 2.4.2 the routing protocols.

2.4.1 Data Routing Issues

Data routing in the IoT is a very important subject, in fact, its importance can never be overemphasised. This is due to the only fact that data is always the centre and most important in almost every field or company or organisation. Usually, questions are asked about the data, it may be routed for sending the data, how the data will be sent, availability, data integrity, security and privacy. Every of these reasons is very vital in its own right; there have been many forms of studies on all of them.

Whenever data is involved in any process, three questions always pop up - data integrity, security and privacy. These always come up because people are concerned about the authenticity of the data they get, how secure it is and also they do not want their data shared with someone else. Another vital question also asked always especially by big companies is availability, everyone needs to get whatever data they need instantaneously.

In order to answer or try to answer these questions and many others that have been or are being asked in IoT about routing, researchers have done great work in providing some routing protocols for the IoT. Also, protocols in other fields such as wireless sensor networks (WSNs) have been used in IoT to solve routing issues. Within the subsequent section, protocols for routing in the IoT will be discussed. A number of these are RPL, Naive routing protocol, Probabilistic routing protocol, 6LoWPAN, e.t.c.

2.4.2 Routing Protocols and Techniques in IoT

There are many routing protocols currently available, some are standard for the IoT while others are not. The standard protocols are those built mainly for IoT while the non-standard protocols are used for other applications such as WSN, some of these have been implemented on different IoT environments. In this section, routing protocols used for Wireless Sensor Networks, IoT and AdHoc Wireless networks will be discussed in detail.

There exists different classifications or categories of routing protocols based on protocol operation or functionality and network structure 25. Routing protocols can be categorised as reactive or proactive protocols. In reactive routing the protocol only looks for a route to a destination when needed - it is also called on-demand routing. In proactive routing, periodic messages are used to send messages to nodes about its neighbourhood and as such is likely to have a route to destinations always. In this section, routing protocols, as well as the different classifications, will be discussed.

2.4.2.1 Routing Protocol for Low-Power and Lossy Networks (RPL)

RPL is a distance-vector which is based on IPv6 and is independent of the link which is used for routing, it is also a source routing protocol 31 32. It was made for low- power and lossy networks and in 2011 it was standardised by IETF 57. It is most often considered the de-factor routing protocol for the IoT.

As already discussed briefly, RPL is said to be made for low-power and lossy networks (LLN), so what then are LLNs? To understand the concept of RPL, this question has to be answered. Another question which needs to be addressed before further discussion on RPL is what distance-vector is and what source routing means also.

A protocol is said to be a distance-vector if it's nodes have the ability to manipulate vectors or arrays of distances to other nodes in the network.This means that the nodes in the protocol do have intra-domain interaction between them. To have an effective interaction between the nodes, there is a need for minimum complexity in computations and message overhead, also each node must inform other nodes (neighbours) of any change in topology 34. The network topology is the pattern of arrangement in which nodes are connected in a network. A distance vector protocol always calculates the direction (address of the next hop) and distance (cost to reach a node) to any node in the network. Every node keeps a vector of the minimum distance (route with the smallest cost) to every node.

LLN is the type of constrained-node network. A constrained-node network is a network which is made of nodes that have some limitations. “LLN: Low-Power and Lossy Network. Normally created from many embedded devices with constrained power, memory, and processing resources interconnected by a variety of links, such as IEEE 802.15.4 or low-power Wi-Fi. There is a wide scope of application areas for LLNs, including industrial monitoring, building automation (heating, ventilation, and air conditioning (HVAC), lighting, access control, fire), connected home, health care, environmental monitoring, urban sensor networks, energy management, assets tracking, and refrigeration.” - RFC 7228

Abbildung in dieser Leseprobe nicht enthalten

Figure 2.10: A DAG, root node and leaf nodes

RPL is discussed earlier is a distance-vector routing protocol which utilises source routing and is the de-facto standard for the IoT. RPL organises the nodes in the network in a topology as a graph called the Direct Acyclic Graph (DAG). An example of a DAG is shown in Figure 2.10. The DAG is then divided into one or more Destination Oriented Graphs (DODAG). A DODAG is a directed graph made of leaf nodes without cycles which are directed towards a single root node 30 33. Figure 2.11 shows a DODAG. Every traffic from each leaf node within the topology is being routed to the root node through only one route. Each node keeps more than one parents to the root (route) but a chosen one is preferred for upward forwarding of data packets to the root node; the other routes are kept as backups 30.

Abbildung in dieser Leseprobe nicht enthalten

Figure 2.11: A DODAG with all leaf nodes directed towards S (root)

In RPL, the root node initiates the network topology by sending messages containing control packages called DODAG Information objects (DIO). The DIO contains information about the graph or network. The leaf nodes receive these messages and process them. After processing the DIO, the nodes decide by using some rules whether to join the network or not. The nodes also use the rules and information to decide which among its neighbours will be its parent node - this might be the root or the parent of a small DODAG. During communication each node sends a Destination Attachment Object (DAO) to its parent if it needs to send data to another node, the parent will process the DAO and transfer the required packets to the destination. The RPL supports three types of communication between nodes - multipoint-to-point, point-to-multipoint and point-to-point.

2.4.2.2 Cognitive RPL (CORPL)

Cognitive and opportunistic RPL - CORPL is an extension of the RPL designed for cognitive networks - a network with a perceptive process which has the ability to observe present conditions of the network, act based on the condition and self-learn from the result of its actions 39. Just like RPL, CORPL uses DODAGs but with some modifications.

In studies carried out in 36, there are some problems that are encountered while using the RPL in cognitive networks, which is supposed to be the default routing protocol IoT. In order to tackle those issues, an improved variant of RPL - CORPL was built for cognitive networks.

CORPL was made to use the DAG just like the RPL, but with an opportunistic approach. In CORPL, there are two major steps; these are: selection of a forwarder set and the unique forwarder selection. In the first step, every node in the network will select as many next hop neighbours as possible. In the second step, the nodes determine the best receiver among the selected forwarder set, using a coordination scheme; when the best receiver is determined, the node will then allow it to forward the relevant packets. Each node keeps a forwarder set from which the next hop/forwarding node is selected opportunistically. According to 37, the use of opportunistic forwarding in CORPL improves end-to-end throughput and reliability; this is achieved by making use of the inbuilt properties of the wireless channel - this is actually a big concern in lossy networks.

2.4.2.3 Lightweight On-demand Ad hoc Distance Vector Routing - next generation (LOADng)

LOADng is a reactive protocol which was designed to provide efficiency, scalability and security in routing in LLNs; It is a distance-vector protocol which is lightweight. It does not keep a routing table for different nodes but works on-demand, it initiates a route discovery when ever there is a need to transfer packets to a destination node; as explained in earlier sections, reactive protocols experience reduced routing overhead and memory consumption compared to the proactive ones 49.

Just like other reactive routing protocols, LOADng has three different messages with which it works, these are Route Request (RREQ), Route Reply (RREP) and Route Error (RERR). The sender node sends RREQ when there is a need for packet delivery in order to discover a path to the destination node; the destination node will send back an RREP after receiving the RREQ from the sender. When there is a link failure/break, the destination sends back RRER to the original sender of the packets it is receiving.

As mentioned earlier, the LOADng is lightweight. The designers of this protocol built it by using minimal core and a small set of protocol operations, also it was built with simple implementation requirements thereby making the code footprint small and the operation state requirements as well. LOADng is quite different from its predecessors, its characteristics are discussed vividly below:

- modular design: this is simply the lightweight core of the protocol. The core makes the protocol extensible with a packet format that is flexible.
- flexible addressing: length of address from 1 to 16 are supported. The only requirement is that all addresses inside a given routing domain of the network must be of the same length.
- Metrics: there is a support for many metric types other than simple hop-count.
- destination-replies: intermediate routers are not allowed to respond to RREQs, only the destination node is allowed, this reduces the complexity of operations, reduce the size of messages and improve security.

LOADng involves three protocol operations - route discovery, path maintenance and path metrics.

2.4.2.4 Collection Tree Protocol (CTP)

This routing protocol was developed primarily for WSNs 31. It is a distance - vector routing algorithm. Before the RPL was developed, CTP was the de-factor routing standard for the TinyOS. It is widely considered as a general reference protocol for WSNs 22 21.

CTP constructs and maintains a tree-based topology using routing messages also called beacons, this reports the data messages to the sink which is the root of the network. To ensure that routing messages are sent to the root, CTP uses the adaptive beaconing mechanism.

2.4.2.5 Channel-Aware Routing Protocol (CARP)

CARP is an underwater wireless sensor network routing protocol. It employs the use of multi-hop data delivery to the sink of the underwater WSN 31. It is a cross layer protocol which takes advantage of link quality information to determine the cross layer delay 50. Using the information pertaining link quality, CARP selects nodes which have up to date history of successful transmissions to their neighbours. The protocol combines the link quality with the hop count, which is the simple topology information to be able to connectivity voids and shadow zones. It is also able to select robust links by taking advantage of power control.

At the start of network setup, the sink (root node) sends a HELLO broadcast message to every node inside the network. With the broadcast message, every node is able to get its hop count - distance to the node, which is very necessary. Using PING and PONG messages, whenever there is any packet that needs to be transferred, the sender node selects the most suitable relay to the destination node. PING is the message sent by the node to initiate a transfer of packets and PONG is the message sent by any node which receives the PING message, that node forms a relay to the destination node 50.

During the exchange of the PING and PONG messages to get a relay, time is recorded. In addition to the time, goodness is computed for each node, the goodness value is then used to calculate the link quality of all possible relays to the destination node. To transmit the packets, the relay with the best link quality is chosen to transfer the packets. While sending the PING messages initially, the power used to send them is also computed; this enables CARP to take advantage of power control to select robust links for packet transfer 50.

2.4.2.6 E-CARP

This protocol is an enhancement on the CARP to support greedy and location-free hop-to-hop routing to ensure energy efficient forwarding of packets from the sensor nodes to the sink. In CARP data acquired by the sensor nodes are not being neglected, with their presence in the network; sometimes unwanted forwarding might be done from those nodes in the network, E-CARP is built to solve this problem by enabling caching of sensory node data at the sink 31.

Another feature that is not being exploited in CARP is reusability of relays in the network. In situations where the network is steady, there is usually no need for a PING-PONG message transfer between nodes. E-CARP is built to exploit reusability of previous links by giving previously used links a high priority before initiating a transfer 31.

[...]