Integrating ECC And ML Into MQTT Protocol For Secure Communication Of IOT Devices

The rapid research advancement in the field of networking has brought a new type of networks,the so-called Internet of Things namely Io T,which is undoubtedly very trending in networking nowadays and lots of research has been done and Various architectures of Io T were designed according to the requirements of the system.The main idea of Io T was the flow of data between devices and ease of communication across a large number of nodes.In the very beginning,communication was a barrier since most of the devices and technologies were still evolving and limited to wired communications.Later wireless technologies like Bluetooth and WI-Fi have made the communication very fast and easy to implement.Io T is a network of physical objects or things that are embedded with electronics,software,sensors,and network connectivity,which enable the objects to collect and exchange data.Rapid proliferation of Io T is driving the intelligence in things used daily in homes,workplaces and industry.An Io T ecosystem consists of web-enabled smart devices that use embedded systems,such as processors,sensors and communication hardware,to collect,send and act on data they acquire from their environments.Io T devices share the sensor data they collect by connecting to an Io T gateway or other edge device where data is either sent to the cloud to be analysed or analysed locally.Sometimes,these devices communicate with other related devices and act on the information they get from one another.The devices do most of the work without human intervention,although people can interact with the devices,for instance,to set them up,give them instructions or access the data.The Internet of Things(Io T)is recognized as one of the most important areas of future technologies and is gaining vast recognition in connecting anyone with anything anywhere anytime and progressively becoming the important aspect of our life by changing the traditional way of living into a high tech life style.Smart transportation,smart city,smart home,pollution control,smart waste management,smart health are some of the transformations due to Io T technology.In order to achieve these desired transformations,researchers developed and deployed a broad variety of Io T application systems using a set of rules,standards and protocols,which reduce the essentials of the implementation of Io T applications.The devices in Io T systems must communicate with each other,the servers must collect the data from these devices,and the data is then analysed or provided to the people.For all this to happen,there is a need for efficient protocols to ensure that the communication is secure and to avoid loss of data.Io T devices communicate using Io T protocols.Internet protocol is a set of rules that dictates how data is sent to the internet.Io T protocols ensure that information from one device or sensor is read and understood by another device,a gateway or a service.Different Io T protocols have been designed and optimized for different scenarios and usage.Given the diverse array of Io T devices available,using the right protocol in the right context is important.Network protocols and standards are policies that comprise certain rules that define communication between two or more devices over the network.Single devices like smartphones and personal computers also use network protocols for communication.However,general protocols that are used by these personal devices might not meet specific requirements like bandwidth,latency,and range of Io T-based solutions.Hence,a few improved versions of existing protocols and some new Io T protocols are evolved to meet the requirements of Io T devices.Io T devices on the network have to agree upon various aspects of data exchange methods for successful transmission of data to happen.There are important protocols,especially Message Queue Telemetry Transport protocol(MQTT)required to establish Device-to-Device communication in Io T environment.Io T protocols are an integral part of the Io T technology stack.Without Io T protocols and standards,hardware would be deemed useless.This is because Io T protocols enable hardware to exchange data.And,out of these transferred pieces of data,useful information can be extracted by the end-user.Using the right protocol according to the environment and the devices in use has a significant importance.This could boost the security of Io T system by a consistent level.Some of the widely used Io T protocols are: Hyper Text Transfer Protocol(HTTP),Constrain Application Protocol(CoAP),Universal plug and play(UPnP),Extensible Messaging and Presence Protocol(XMPP),Data Distribution Service(DDS),Message Queuing Telemetry Transport(MQTT)and so on.In general,all various Io T protocols adopt two main messaging patterns,namely:Publish/Subscribe and Request/Response.However,publish/subscribe messaging pattern has more advantages than the request/response pattern for resource-constrained devices.Request/response protocols such as HTTP are typically resource demanding.In addition,the RESTful mode used in CoAP depends on regular updates or polling,which can prove to be resource intensive.Universal Plug and Play(UPnP)is a discovery protocol for Io T devices to discover and connect with other compatible devices.The new version of UPnP protocol released by the UPnP group in 2014 uses the XMPP protocol,to connect UPnP devices as XMPP clients.The owner of XMPP protocol revised it to support both request/response and publish/subscribe message patterns with greater support for Io T deployments.However,there exist security challenges with the use of XMPP protocol in the Io T deployments as highlighted by in multiple researches.The Publish/Subscribe concept of MQTT protocol allows for mass distribution of information to interested parties in an efficient manner.It reduces network traffic by up to half,by granting the publisher of data to send its data only once to a publish/subscribe server,who then redeliver it to subscribers.The publish/subscribe pattern(also known as pub/sub)come up with an alternative to traditional client-server architecture.In the client-server model,a client communicates directly with an endpoint.The pub/sub concept decouples the client that sends a message(the publisher)from the client that receives the messages(the subscribers).The publishers and subscribers never contact each other directly.In fact,they are not even aware that the other party exists.Among various Io T protocols,MQTT is easy to implement and extremely lightweight designed for low bandwidth,unreliable network and low power consumption,which make it suitable for Io T devices communication.MQTT operates on top of TCP and was not designed with security in mind but wanting to achieve a very simple and light protocol to minimize bandwidth and energy usage.Thus,security in MQTT is a major problem in terms of authentication,authorization,confidentiality and integrity.Regarding the lack of confidentiality present in MQTT we can say that packets in MQTT can be spied by an attacker because no encryption is applied.Thus a secure mechanism is needed over the transport protocol.Some works have been done in this area and more research is on-going to achieve a secure communication using TLS/SSL on top of MQTT or adding extra layers of security.Security challenges of MQTT protocol rely on the way that it is implemented,configured and certain considerations like not having security built in and protocol having concepts such as wild-card topics and linked resources can be turned against users by exposing their resources and collecting data about them.In addition,Data privacy in MQTT message is absolutely an issue since,by default,MQTT does not provide any data encryption.Whether the broker system uses authentication mechanism or not,the attacker can still sniff the data in transit easily.Moreover,MQTT does not check the data or payload that they transport.To overcome these challenges,we have to develop protocols that are able to analyse the traffic and secure the data being transmitted over the Io T system.Security in Io T networks is the main concern for developers.Since most Io T nodes are limited in resources,they require a security mechanism that fits their limitations.This research provides a lightweight security mechanism for Io T resource-constrained devices over the MQTT protocol.The Contributions of this thesis are based on the development and deployment of a Secured Message Queuing Telemetry Transport(MQTT)protocol based on lightweight Elliptic Curve Cryptography(ECC)and machine learning(ML)algorithm for Io T Infrastructure.For the broker to handle the processing power required by machine learning algorithms,we suggest to use a powerful Io T device with mobile GPU for the broker node,or a normal computer as the controller of the system namely “broker”.ECC is an asymmetric key algorithm that is using public key encryption technique to create faster,smaller,efficient and well-structured cryptographic keys.Asymmetric key algorithm uses pair of keys: public key and private key to complete the message encryption and decryption process.If the message is encrypted using a public key from the pair,then it is only possible to decrypt using the other private key of the same pair.Instead of using some traditional methods,ECC uses the properties of the elliptic curve equation to generate keys.The logic idea of ECC is that the key we use to encrypt our data can be made public while the key that is used to decrypt our data can be kept private.What we need in an asymmetric cryptographic system to work is a set of algorithms that is easy to process in one direction,but difficult in return.The results of this thesis are showing that ECC is an effective approach to secure the data on MQTT protocol against unauthorized access.Even if data is secured against an authorized access by using encryption mechanism provided by Elliptic Curve Cryptography,attacker could remain on the network and launch a denial of service attack or any other kind of attack to the broker by sending many connection requests continuously,thereby making the broker busy as in flooding attack.As there are many connection requests that arrive at the same time then the broker will be drained and not be in a very good position handle new incoming requests.Moreover,the broker is not able to differentiate the legitimate or normal connect and the spoofed connect message packets.By receiving the flood request messages,the broker starts to acknowledge with connect acknowledgement message.There is an abrupt rate of increase in the number of connect and connect acknowledgement during the denial of service attack.This results in blocking the broker service and prevents the functionality of the intended Io T network.It is from this scenario where we have fetched the idea of implementing some Machine Learning model in order to identify and predict the type of attack with the aim of misbehaving,blocking or slowing down the performance of our protocol.Machine learning(ML)is a field of study in computer technology that allows computing devices to perform a task by learning from data(i.e.,gradually improving performance with experience)without having to be explicitly programmed.It gives to the system the ability to learn from data,identify patterns and make decision with minimal human intervention.Machine learning models can predict the expected behavior of a system by learning from previous experiences.Therefore,applying ML models can significantly advance the security methods by transforming the security of Io T systems from simply secure communication between devices to security-based intelligence systems.Many studies have shown that among several ML algorithms,Artificial Neural Networks(ANN)has the ability to learn and model non-linear and complex relationships,which is important because in real-life,many of the relationships between inputs and outputs are non-linear as well as complex.Furthermore,unlike many other prediction techniques,ANN does not impose any restrictions on the input variables.By implementing and analysing different machine learning algorithms,we have found out that ANN outperforms other algorithms in comparison for identification and prediction of the type of attack with accuracy of 0.9038%.In this thesis,we represented the historical background and some previous works related to our subject.Moreover,we described the methodology and the system models that have been applied in this thesis and their implementations,as well as the analysis of the obtained results in different forms.We ended up by making conclusion and recommendation of future motivation works.