section{The computing as latter is built around events between

section{The Serverless architecture and Internet of Things}label{sec:CM} Nowadays most of the current trend applications consist of two main components. Interface that runs on the user’s device (UI for website and application) and the server side which processes the request from user.\    Whereas this paradigm is not applicable to the Internet of Things application. Due to in the traditional Internet of Things and cloud architecture, the data produced by the devices transport to the cloud for further processing such as calculating, store or analysis. But the Internet of Things is about to add millions of new devices which is similar to the million of User Interface interact to the network system which risks overwhelming of the request handler for the network infrastructure. Currently, the solution to this solve this problem is push the data processing part closer to the devices which known as Edge computing and Fog computingcite{EdgeFog}.\subsubsection{Edge computing}Edge computing is a first step of the IoT and cloud architecture. The concept is pushing the processing power and communication capabilities of an edge gateway from the server direct to devices. Edge computing has an ability to collect, analyze, and process data from the physical assets or sensors that connected with the system.cite{EdgeFog}subsubsection{Fog computing}Fog computing is an architecture that uses one or more collaborative clients (fog node or Internet of Things gateway) to carry out a substantial amount of storage, communication, control, configuration, measurement and management. Fog computing, transporting data from things to the cloud considered as an extension of the cloud computing paradigm from the core of network to the edge of the networkcite{EdgeFog}cite{FogConcept}.subsection{A role of Serverless in Internet of Things economy} There is a common relevance of FaaS and the edge computing as latter is built around events between device and fog then do the post-processing in fog and cloud. The example is in IoT architectures the sensors such as sensors which collect the data within a period interval of time. For example in industry we have wind pressure sensors, temperature sensors, and in the daily life we have smart phone sensors, such as noise detection sensors, vibration measuring sensorscite{FogConcept}. Therefore Serverless architecture which bases on event-driven strongly related to the Internet of Thing approach, furthermore in the fact that hardware abstraction is a key of an ecosystem around the Internet of Thing device, Serverless architecture allows a user to scale without the hardware limitation. It also provides possibilities for a developer to write and deploy code, regardless of the specifics of particular devices, that is executed on devices.cite{Kappa}\subsection{Existing technologies related with IOT and Serverless architecture : Amazon Web Services}Amazon web services is an on-demand cloud platform which offering tools for implementation of the web server, storages, services like content delivery and other functionalities to drive the application. One such service is AWS IoT, AWS Greengrass and AWS Lambda which already described in the previous section.\    In the past, AWS Lambda user can only use Lambda function via accessing the cloud but which introduce of AWS Greengrass user can use AWS Lambda at the local or at the edge of the system and connect to IoT devices by AWS IoT. This means Lambda function can run mostly everywhere- house, the car even television. With this fact AWS becoming the dominant factor in the IoT and cloud system. subsubsection{AWS IoT} AWS IoT 5 is a platform to integrate Amazon Web Services for IoT applications. AWS IoT allow us to connect and manage devices in our system securely, Without any effort the platform also allows us to connect devices into the cloud, organize collect, store, analyze and take action against the increasing volumes of data from connected devices.\AWS IoT consists of following components: \1. Device gateway: This part enables devices to securely and efficiently communicate with AWS IoT.\2. Message Broker: This part provides mechanism for devices and AWS IoT applications to publish and subscribe message from each other.\3. Rules Engine: This part provides integration of messages to other services like Amazon S3, Amazon Greengrass, Amazon DynamoDB and Amazon Lambda.\4.Security and Identity service : Automatically imply from naming this part provides security and authorization for devices\5. Thing Registry: This part provides the registry for organizing the resources.\6. Thing Shadow: This part provides ability to create background task. For example, JSON document which used for storing and retrieving the current state of information.\subsubsection{AWS Greengrass} AWS Greengrass6 is software which responsible for connecting AWS Lambda and AWS IoT functionality together and makes it available for local deployment and execution on the local devices that occurring at irregular intervals network connections, making it possible for them to collect and analyze data closer to the source of information and also provide the ability to run local computation, messaging, data caching and with Machine Learning inference embed in the system user can easily create their own Machine Learning model.More generally, developers can run Serverless code (from AWS Lambda) in the cloud and conveniently deploy it to devices for local execution of applications.\AWS Greengrass consists of following components:\1.AWS Greengrass core software: This part bring the Lambda functions to user local device, user can deploy and execute easily in the local machine.\2.AWS Greengrass core SDK: This part allows Lambda functions in local machine able run the Lambda in the runtime.\3.Cloud service: This part provides the connection between AWS cloud and local application.\4.Others extended Features such as Lambda runtime, Thing shadows implementation, Message manager, Group management and Discovery service.\ egin{figure}!htb includegraphicswidth=0.49 extwidth{images/GreengrassArchitecture} caption{Basic Architecture of AWS Greengrass (from cite{architecture})} label{fig:TCPIP}end{figure}\Benefit of AWS Greengrasscite{AWSGreengrass}:\1. Respond to Local Events in Near Real-time: AWS Greengrass devices can interact with the data locally if local computing power is not enough. AWS Greeengrass able to use the cloud for management, analytics and store the data. The local access feature allows Lambda functions on Greengrass Core to use device local resources like cameras, sound, CPUs or GPUs.\2.Operate Offline: AWS Greengrass synchronizes the data from cloud to the device.\3.Secure Communication: AWS Greengrass always authenticates and encrypts the device data for both local and cloud communications.\4.Simplified Device Programming with AWS Lambda: AWS Greengrass and AWS Lambda use the same programming model, so you can develop code in the cloud and then deploy it to your local devices.\5.Reduce the Cost of Running IoT Applications:  AWS Greengrass let your device filter the data locally with this ability user do not have to upload not necessary data into the cloud.\subsection{The example work of Serverless architecture using in Internet of Things} The term of the Internet of Things can apply to various application domains in our life such as sport, industry, daily life and even can be apart to another subject such as algorithm or data science. Connect Internet of Things with the cloud platform provide huge computing power and almost unlimited storages. This section will show the example of the Internet of Things application from the different domain of Internet of Things using Serverless cloud base approach. What is the reason they decided to use Serverless cloud architecture apart from their application and what is the architecture they use to connect to Serverless cloud and IoT devices?   subsubsection{Serverless computing for smart grid architecture4} A smart grid generates a lot of data, these data need to be stored, processed and managed properly so it necessary to have a proper method which can trigger the control system of the smart grid whenever necessary.paragraph{Targeted system}     Propose the additional digital layer which gathering the data from the smart grid architecture using Serverless cloud computing framework and cloud database. First, the application packed the generated data into a file and upload to the cloud then the control algorithm will check into the data whether the file contains problematic data or not and output the result as a file.\paragraph{Implementation} Figure xxx show the over view of the system implementation started with Lambda function and s3 bucket. AWS s3 is the scalable cloud storage and Lambda is the aspect which code will be executed and interact with s3 or Grid model event.And a MATLAB program used to generate the text file which used as batch program to upload the data to s3 and trigger the lambda function. egin{figure}!htb includegraphicswidth=0.49 extwidth{images/smartGrid} caption{Illustration of Grid model interaction with cloud platform (from cite{architecture})} label{fig:TCPIP} end{figure}paragraph{Advantage of Serverless state in this paper} The cost of Serverless architecture is cheaper than installing the servers in the cloud. For example, for a single Linux instance in EC2 will cost 0.312 US-dollar per day, Changing the CPU capacity add more further make it hard to scale the system. But for the AWS Lambda in the same amount of time (average 10,000 requests per day) will cost just around 0.002 US-dollar and 0.00001667 US-dollar per GB-second response time. Moreover, Serverless architecture is easier to implement the protection algorithm and extend the storage.XXXX.    subsubsection{A Serverless Real-Time Data Analytics Platform for Edge Computing} Growing on the Internet of Things. Edge computing produces a lot of data. But a lot of them still remain inaccessible because of a various problem such as networking cost, latency issues and limited interoperability between edge devices. The reason is cloud models do not fully support data analytics at the huge volume and variety type of the data produced by many sensors. paragraph{Targeted system} To propose a full-stack cloud and edge data analytics platform supporting real-time data analytics, using Serverless architecture execution model with programmatic resources and data analytics management. Which enable users freeing from worrying about the complexity of the underlying edge infrastructure.  paragraph{Implementation} Figure xxxxxa shows the architecture ability to manage, collect and analyze the data from a various source. The edge focused on local (hosting edge gateway) while cloud will support global, combining and analyzing data from different edge, regions or even domains(Not IoT system). So the data analytics can be performed on edge nodes and cloud nodes. Allows developer to define the analytics function and data-processing business logic and application goals without dealing with data transform management, orchestration and optimization.\       Figure xxxxb shows the core of the platform which dealing with input data handling, real-time analytics functions and the others function for support the system such as deployment, quality of service, elasticity and else.\paragraph{Advantage of Serverless state in this paper}     The Serverless model enables easier and more intuitive development of real-time analytics functions without worrying about nonfunctional requirements such as the scaling of the system, heterogeneous data management, routing and response time.egin{figure}!htb includegraphicswidth=0.49 extwidth{images/dataAnalytic} caption{Cloud and edge real-time data analytics platform (a) High-level usage context (b)Internal software architecture. (from cite{architecture})} label{fig:TCPIP} end{figure}subsubsection{Experiences Creating a Framework for Smart Traffic Control using AWS IOT} The propose of this application is creating dynamic vehicle traffic control based on the number of vehicles and the schedule of public transport such as an underground train. The primary challenges of the implementation are application manageability and design to reach real-timeliness with latency performance, asynchronicity, and scalability.    This paper addresses the challenge of implementing a scalable IoT infrastructure which they call testbed in the public cloud for scientific experimentation. There are two main contributions of this paper. Firstly, the design and implementation of a Serverless cloud-based IoT infrastructure in AWS. The system is creating for dynamic vehicle traffic control. Specifically, in the road sensors, and vehicle GPS positioning comprise the input to a platform to calculate the red-green patterns of traffic lights, the goal of this scenario is securing the safety and minimizing the waiting times. Secondly, the system should be able to handle the even in real-time for both stateless and stateful. Specifically, any sensors can send the data to the cloud without modifying additional control logic. paragraph{Targeted system} The system needs to be able to orchestrate both physical objects and virtual objects with scalability in real-time. A system state shall be able to process administrative actions as well as the states of the system should be able to be triggered by physical devices and vehicles GPS and also able to investigate or debug the flow of the data easily in real-time. The system needs to be able to scale to a large number of devices without editing logical control algorithm and not affecting the real-timeliness nor limit the number of concurrent data processes. And also the system designed for a scientific team, therefore, they do not expect to have to develop and maintain further application infrastructure.\ paragraph{Implementation} The implementation using AWS components is presented in the figurexxx. AWS was chosen because in the current period AWS is the most comprehensive portfolio to construct a scalable, extensible cloud IoT infrastructure. Moreover, the AWS cloud provided the architecture interacting with both virtual resource which could be any application and the actual physical IoT devices. following is the application was used to construct the real-time vehicle control system.\1.AWS IoT: As described in the previous section, AWS IoT offers data stream endpoint bridge connection and message routing with a simply stateless rule engine.2.AWS Lambda: As described in the previous section, AWS Lambda function is a state-less programmatic base on the event from an input and an output. Lambda function can be a call from both internal local system and external AWS system via endpoint access.3.DynamoDB: DynamoDB is a schema based No-SQL DB. In this work, AWS DynamoDB is used for storing collected data from any source and maintaining states of the application.4.Kinesis: Kinesis is a highly scalable aggregating streaming data buffer. Can achieve a high throughput by forwarding the data to an infinite pool of parallel end-points. Kinesis also has an ability for maintaining a maximum throughput and ensuring the endpoints will be up to date in a period of time this approach securing the ability of the system to reach real-time system.5.CloudWatch : CloudWatch is a platform for monitoring the logs file of AWS services belongs with predefined and custom metrics.\ egin{figure}!htb includegraphicswidth=0.49 extwidth{images/traffic} caption{AWS components and data flow. (from cite{architecture})} label{fig:TCPIP} end{figure}\paragraph{Advantage of Serverless state in this paper} Serverless automatically scales to match with the growing number of the sensor. This ensures that the information will flow from end-to-end in almost in real-time without the hardware optimizing from developers.    There are numbers of the benefit from an AWS Lambda function (FaaS) as compared to an EC2 instance (IaaS). For the AWS Lambda, the system is based on event-driven, The container does not appear all the time. But for EC2 an equivalent EC2 instance have to run continuously all the time, and the software has to be able to scale to multiple evaluations while guaranteeing real-timeliness on a machine. with Lambda functions the system able to reach scalability and real-timeliness requirements without any further implement or setup from the developer.