IoT : Cloud @ Edge

In the previous blog IoT: Interfacing physical realm with digital space we were introduced to a new digital age which will usher a community of digital beings. The humans and machines will be part of an interconnected mesh of billion of digital entities.

The new connected ecosystem of smart machines handling critical functionality of our lives poses challenges in networking, security and computing. In this blog, we will uncover cloud and edge computing and how they help in addressing challenges of IoT. The design principles mentioned in this blog are artifacts of my experiments with IoT.

Design Principles :
The fusion of cloud with edge devices should adhere to certain design principles in order to build a securely connected ecosystem, using distributed computing and utilizing network optimally to reduce the cost of operation and improve productivity.
1. Cloud Computing :
IoT uses the power of the cloud for costly operations like data mining and predictive analytics, upstream and downstream messaging. The machines send the data captured by sensors to cloud services deployed on the internet. The machines directly send and receive messages from and to the cloud. The evolution of embedded computing not only allows to monitor devices using sensor information but also control them using actuators. The machines are thus exposed to security threat as they are directly connected to the insecure mesh of internet.

The latency in communication from cloud to the machine makes the machines less responsive to change in their local environment. There are situations where the machines have to be quick in decision making and respond to critical change in local environment spontaneously

2. Edge Computing :
Edge computing is localizing certain kind of analysis and decision making capabilities enabling quicker response time, less affected by network latency, reduced traffic and distributing intelligence between cloud and edge devices. It addresses networking security and computational challenge. Thus it helps in realizing the full computational potential of the complete connected platform.

3. Connection :
Devices in the local ecosystem are connected to each other using WSN(Wireless Sensor Networks) which are not exposed to the commercial internet. The gateway of local environment creates an ad hoc wireless network which connects all edge devices(sensors, actuators, etc). The gateway also acts as an access point for global internet access to edge devices. The gateway deploys a firewall for all communication with external internet services. This insulates the local ecosystem from the global internet and secures inter-device communication from external attacks. This design also decentralizes the network topology to handle single point of failure.

4. Integration :
Integrating embedded devices, sensors, software applications, external services and cloud services using resilient messaging infrastructure is critical for operational efficiency of an IoT solution. Evolving communication infrastructure is pushing the boundaries of IoT.

There are short-range communication networks like DSRC(Dedicated Short Range Communication) which is a short to medium range communication channel specially designed for machine to machine communication. It comes with its own set of protocols and operates at 5.9 GHz band exclusively dedicated to it. This type of networks creates WSN(Wireless Sensor Networks) for connecting entities in the local ecosystem.

LPWAN(Low Power Wide Area Network) like NarrowBand-IoT, 5G is new age radio technologies designed for long-range cellular communications and low power consumption. The low power communication comes as a blessing for embedded devices with limited power capacity.

5. Messaging :
The messaging infrastructure is the backbone of any IoT solution. The messaging framework should have certain attributes in a connected solution ;

Lightweight – IoT solutions have myriad of digital entities connected in a mesh. In order to reduce the load on traffic, the packets should be lightweight. Embedded devices with limited capacity can afford messages with small code footprint for best performance.
Local and Global Communication – Deploying a WSN for communication in local environment reduces long distance communication with the cloud, thereby reducing latency and network traffic.
QoS(Quality of Service) – QoS is a technique to handle network requirements and manages resources. IoT solution encompasses a wide range of communication types ranging from guranteed device to device call, the device to message broker, cloud to user devices and vice-versa. QoS tailors the network requirements by managing traffic delay, jitter, packet loss and bandwidth. This ensures that the network is used optimally.
Security – Wide range of communications makes the system vulnerable to attacks. The criticality of functionality and sensitivity of data shared among entities demands best security practices to mitigate risks.

6. Security :
The security considerations for IoT solution comprises of secure communication, robust embedded platforms, insulation of environments. There is still a long way to go in building a secure infrastructure for IoT. Some principles at the core of building security framework are :
Isolating environments – In order to ensure mitigation from external attacks we have to insulate local environment from global ecosystem. Connecting the edge devices using secure WSN is essential for intra-device communication. The gateway acts as the only access point to the external internet. We deploy a firewall at the gateway for filtering messages.
PKI(Public Key Infrastructure) – Setting up a PKI is essential for long-range communication between entities like the gateway, cloud, client devices, backend servers. PKI ensures secure electronic communication by managing public key encryption and distributing, storing, creating identity certificates also known as digital certificates or public key certificate.
Virtualization – In a smart city or huge production plant embedded devices are located remotely like on highways or tunnels. In case of system crashes rebooting remotely is not possible and it is a cumbersome task to repair it physically. Virtualization allows hosting of a number of VMs(Virtual Machine) on same SoC(System on Chip) completely sandboxed from each other thereby ensuring that mission-critical services continue to operate even if one particular component fails. For example, if JVM crashes it has no effect on critical Linux virtual machine running on the same chip.

Use Case – IoT in Manufacturing :
IoT is evolving into a disruptive force in all sectors including automotive, manufacturing, healthcare. In this blog, we will understand the current problem statements of manufacturing industry and how IoT has potential answers to those problems.

Problem Statement: In order to reduce the operational cost and maximize the operational efficiency, manufacturing industry has identified these improvisation areas :
Gain end to end visibility across entire production process.
Connect production to the core business.
Build responsive manufacturing to meet customer needs.
Real-time asset health monitoring
Predictive maintenance.
Logistics and Supply chain
Solution Overview: In order to use IoT to answer the above problem statement I have implemented my design principles for crafting a solution for smart manufacturing :
Connecting production units with operations and business systems using embedded devices and software applications.
Real-time monitoring using sensors for control and optimization.
Message brokers for sending right information to the right person at right time.
Machine learning algorithms to detect pattern and anomalies in the production process.
Predict maintenance and flag machines before downtime or actual error.
Integrating predictive information as continuous data feed with factory’s existing systems and operational process using REST API(producers and consumers)

This is one of the applications of IoT showcasing the power of the cloud and edge computing. With distributed computational tasks and tailor-made network configuration, IoT can provide a disruptive and cost-effective solution in diverse sectors. The weapons of a connected ecosystem, real-time data and decision making based on predictive analysis empowers industries to operate consciously and innovate effectively.