Internet of Things Tutorial: Chapter Two

Internet of Things Tutorial: Chapter Two

In our Internet-of-Things (IoT) introduction we highlighted Wireless Sensor Networks (WSN) and Radio Frequency Identification (RFID) as two of the most prominent IoT technologies. Indeed, these two technologies can be considered as the forerunners of IoT. During the previous decade, IoT was in most people’s minds directly associated with WSN and/or RFID, and we can be sure that there is still is.

Although there is a host of articles, but also dedicated books introducing and presenting these two technologies, I thought there is merit in briefly discussing these them as part of an IoT tutorial. The rationale is that they are key to understanding more complex IoT concepts and technologies, but also the structure of non-trivial IoT systems, which are very likely to comprise RFID or WSN components.

Wireless Sensor Networks
WSN are pervasive deeply networked systems, which compirse/interconnect multiple sensors in a wireless fashion. The motivation behind the creation of WSN systems is that the vast majority of processors are already (and will still be in the future), not in traditional computer systems (e.g., desktops, laptops), but in house-hold appliances, vehicles, machines on factory floors. As already discussed this is a driven for IoT as well. Given this situation, huge opportunities could emerge from adding reliable wireless communications and sensing functions to the billions of embedded computing devices, in order to create ubiquitous networks. In this context, distributed WSN represent collections of embedded sensor devices with networking capabilities.

WSN’s comprise sensors. Nowadays, we are witnessing a proliferation of sensors (including low-cost multi-purpose sensors), based on advanced in Micro Electro Mechanical (MEMS) technology (i.e. very small devices. Sensors comprise an Integrated Wireless Transceiver and have memory, CPU and a battery. Sensors are associated with limitations in terms of energy, computation, storage, transmission range and bandwidth. Their characteristics in terms of the above property define their functionlity, use and cost. A typical WSN system/network comprise numerous sensor node (e.g., hundred to thousand of sensor nodes). Sensor nodes are responsible for collecting information about the environment for and sending it towards a sink node, which receives the information gathered by the network and delivers it to the end-user of the WSN.

In this way, WSN present several advantages over individual sensors, including:

(A) Wider range of sensing, since they cover a more extensive area;
(B) Higher Redundancy, as multiple nodes close to each other can be deployed in order increase fault tolerance;
(C) Better accuracy, given that Sensor nodes can collaborate and combine/fuse their data in order to increase the accuracy of sensed data;
(D) Broader/enhanced functionalities, as sensor nodes can offer forwarding services in addition to sensing functionalities.

Based on these advantages, WSN are deployed and used in the scope of a variety of applications including:

  • Physical security for military operations.
  • Indoor/Outdoor Environmental monitoring.
  • Seismic and structural monitoring.
  • Industrial automation.
  • Bio-medical applications.
  • Health and Wellness Monitoring.
  • Inventory Location Awareness.
  • Future consumer applications

Despite commercial availability of WSN, there is still on-going research on WSN, along with deployment challenges. Some of the challenges include:

(A) The need to deploy deeply distributed architectures, where application logic is distributed across all the different nodes, without centralized goal setting or control;
(B) The need to increase the autonomy of the WSN on the basis of properties such as self-organization, self-configuration, adaptation to the environment, self-healing (e.g., recovery from failures) and more;
(C) Energy convervation;
(D) Scalability in terms of the density of nodes and the number of interconnected WSN;
(E) The ability to deploy data centric networks, supporting adaptation and in-network aggregation of data.

In general, the design and deployment of a WSN, targets the following objectives:

(A) To minimize battery power, storage and computation;
(B) To operate at low bandwidth with the lowest possible error rates;
(C) To be scalable to 1000s’ of nodes;
(D) To operate on the basis of simple, but efficient protocols;
(E) To limite the memory footprint of protocols as much as possible and
(F) To operate in self-configured mode.

A WSN compirse a communication architecture, dealing with data transmission issues, and a middleware architecture which deals with application development and deployment issues. The communications architecture strives to combine power and routing awareness, while at the same time communicating efficiently (e.g., in terms of power/energy) in a wireless fasion. Moreover, the communication architecture facilitates cooperation among sensor nodes. Typically, the communcation part of the WSN comprises:

  • Physical Layer, which addresses the need for simple but robust modulation, transmission, and receiving techniques. At this layer the following are performed: frequency selection, carrier frequency generation, signal detection and propagation, as well as signal modulation and data encryption.
  • Data Link Layer, which is responsible for the multiplexing of data streams, data frame detection, the medium access and error control.
  • Network Layer, which deals with data centric functionalities in order to boost power efficiency. Data aggregation may however be employed, when it is not a set-back to the collaboration of the sensor nodes. At the network layer, addressing based on the attributes and location of each node is implemented (i.e. including location awareness)
  • Transport Layer,which is needed for systems that are accessible through external networks or the Internet. In such case popular protocols (for the transport layer) such as TCP/UDP type can be used.
  • Application Layer, where management protocols are implemented in order to render the hardware and software of the lower layers transparent to the sensor network management applications. At this layer, protocols such as the Sensor management protocol (SMP), the Task assignment and data advertisement protocol (TADAP) and the Sensor query and data dissemination protocol (SQDDP) are implemented.

As a prominent example of an application layer functionality, the SMP implements rules for data aggregation, attribute-based naming, and clustering to the sensor nodes. It also enables data exchange for location awareness, time synchronization of the sensor nodes, algorithms for the mobility of the sensor nodes, functionalities for turning sensor nodes on and off, functionalities for monitoring the status of the network (such as the sensor network configuration), as well as the ever importance authentication and secure data communications functionalities.

WSN middleware architectures (and platforms) provide the software infrastructure needed in order to glue together network hardware, operating systems, network stacks, and applications. The middleware infrastructure provides standardized system services to diverse applications, along with a a runtime environment (container) that can support and coordinate multiple applications. Furthermore, it caters for the adaptive and efficient utilization of system resources. WSN middleware services facilitate development, maintenance, deployment and execution of sensing-based applications.

These services needs to deal with WSN related challenges, including limited power and resources, scalability, mobility, dynamic network topologies, heterogeneity (e.g., in terms of CPU-power, networking, memory and storage, operating systems), data aggregation, security, as well as customization to the needs of the specific application (based on a proper balancing of application requirements with the generality of the middleware functions).
A number of different type of WSN middleware systems have emerged, including:

  • Virtual Machine (Cluster-Based), which exploit the virtual machine concept in order to decouple the application semantics from the underlying physical infrastructure. It has the advantage of providing a common abstraction layer and sand-box for multiple infrastructures. On the downside this approach is associated with high overhead and difficulties in exploiting heterogeneous infrastructures. An example of such a middleware infrastructure is the TinyOS.
  • Sensor Databases, which abstract entire sensor networks as a virtual relational database. This approach facilitates interoperability with legacy systems, but is weak in supporting real-time applications, and in providing very accurate results.
  • Message Oriented Middleware (MOM) Systems, which use publish-subscribe protocols in order to support asynchronous communication between sensor nodes. This allows for a loose coupling between the sender and the receiver. The drawback of MOM techniques are their high overhead. An example of MOM systems for sensor networks is TinyMQ.

Sensors and sensor nodes are an integral part of most IoT systems. In several cases, people are classifying their sensors deployments as IoT. Likewise, people using and deploying WSN middleware platforms such as the Global Sensor Networks (GSN) platform, refer to their applications as IoT. The same holds for RFID, the other forerunner technology for IoT, which is introduced in the following paragraphs. You could access a nice video tutorial on WSN and its applications here.

Radio Frequency Identification

RFID labels/tags can be attached to physical world items and tranform them to elementary pervasive networked sysems/computers (even though not Turing complete machines), which can transmit information to the internet. An RFID system comprises a unique identifier (such as an Electronic Product Code (EPC)) and a Non-volatile random-access memory (NVRAM) storing User Data. An RF system allows wireless exchange of ID/EPC & User Data based on a set of tags, interrogators, as well as controlling Middleware. In particular:

  • RFID Tags have a unique EPC and store a piece of USER Data.
  • RFID Interrogators access USER Data on RFID Tags.

An RFID system can be seen as a distributed database or even a network of distributed tagged objects (i.e. an Intenet-of-Things!).

An RFID system is just a reader and a tag communicating over the air at a certain frequency. It comprises readers, antennas and tags.An RFID solution uses a radio frequency (RF) signal to broadcast the data captured and maintained in an RFID chip. An RFID system is composed of three components: a programmable transponder or tag, a reader (with an antenna), and a host. The operation of the RFID system is as follows:

  • The reader sends out an electromagnetic wave at one specific frequency.
  • That wave hits the RFID tag, and the tag then “scatters back” a wave at a different frequency with the chip’s information encoded in those Backscatter waves.

Note that an RFID reader is a radio that picks up analog signals. The reader not only generates the signal that goes out through the antenna into space, but also listens for a response from the tag. It receives analog waves and then turns them into bits of digital information. Each reader is connected to one or more antennas.

RFID technology is comparable to other automatic identification technologies, such as the uniquitous bar-code technology. There are pros and cons when it comes to evaluating the use of one among those two technologies. RFID is superior in reading range, (data holding) capacity and security, but barcode is still much cheaper, which explains its widespread adoption as the primary automatic identification technology. However, RFID has also the advantage of serialization (nowadays provided however by bar codes), which enables unique iterm identification rather than identification of item classes as in the conventional barcode case. There is a great variety of RFID tags, in terms of functionality, sizes, applications etc. From a functional perspective, RFID tags can be classified in the following categories:

  • Passive Identity Card:Contains only the Electronic Product Code (EPC) in an unalterable form and a CRC for transmission error detection. Also referred to as a “license plate”.
  • Passive Functional Tag: Broad category that includes any tag with functions over and above the elementary tag. Examples of such functions or features include User Writable memory, sensors, and encryption.
  • Semi – Passive Tag: Any tag that embeds battery technology to assist in providing power for the tag (i.e., the battery is not the sole source of energy for the tag).
  • Active Tag: Any tag where a battery is the sole source of energy for the tag.

The benefits of RFID technology include:

  • Access Serialized Datai.e. ability to keep accurate account of items and their properties. This enables one to know what, where, and why.
  • Greate automation, as a result of reduced human intervention i.e. not processing one but many items.
  • Enhance Traceability,which can result in higher – throughput supply chains.
  • Security and prevention of counterfeiting (e.g., in sectors such as apparel and pharmaceuticals where there is a grey Market).
  • Added Value Services, through the use of specialized RFID tags that can sense physical quantities (e.g., location or temperature). Based on such sensors, Location based services and Cold Chain Management services can be deployed.

On the downside of RFID technologies there are significant privacy concerns, which are very well explained in the scope of Ari Juels paper, titled: “RFID Security and Privacy: A Research Survey”.

You can see some real-life RFID applications in practice as part of our RFID-ROI-SME project, and more specifically as part of the YouTube Channel of the project. A short description of the applications can be found in our RFID Journal Article: Toffaletti S., & Soldatos J. (2010, June, 14), RFID-ROI-SME Project Promises Big Help for Small Business, RFID Journal.

View all IoT Tutorial Chapters