Communication is an essential aspect of life for all species, both within (human to human) and between them (human to animal, human to plant, etc.). This communication can concern two actors, one actor and one group, several groups, etc.
We will limit ourselves to communication centered on the transmission of information, as defined in Chapter 1.
Communicating information involves several elements:
We are entering a complex world of hardware, algorithms, software, conventions, etc., and we have to deal with a lot of different things. This world fascinates many research teams, whether in universities or telecommunications industries.
After mentioning a few dates that have marked their development, this chapter will approach computer networks from three angles:
We will finish on an aspect that concerns us all: security.
Computer networks have undergone tremendous development, from the first connection between two machines to cloud computing. We will limit ourselves to the aspects that have structured this development and to those that we may encounter in our activities.
On the basis of what we know today, the first communication systems date back to the 19th century. Here are some key dates:
During World War II, the laboratories of the warring parties perfected new applications:
A computer network is a set of computers that connect to each other and exchange information. In addition to computers, a network can also contain specialized equipment such as modems, hubs, routers and many others that we will discuss in this section.
Computer networks are composed of three elements:
There are three important characteristics of network infrastructures:
Networks can be classified according to their extent, with four very common terms summarizing this extent.
Personal Area Networks (PANs) are restricted networks of computer equipment that are usually for personal use (computer, printer, telephone, etc.).
Local Area Networks (LANs) are mainly intended for local communications, generally within the same entity (company, administration, school, home, etc.), over short distances (a few kilometers maximum). They can connect from two to a few hundred computers with cables or wireless connections. Ethernet LANs are the most common, thanks to the simplicity of their implementation and the gradual increase in connection speeds, from 10 Mbit/s, then 100 Mbit/s, to 1 Gbit/s, then 10 Gbit/s.
Metropolitan Area Networks (MANs) are generally the size of a city and are interconnect networks (LANs or other networks), using dedicated high-speed lines (especially optical fibers).
Wide Area Networks (WANs) interconnect multiple LANs or MANs over large geographic distances, on a national or global scale. The largest WAN is the Internet.
To connect two distant forms of computer equipment, a transmission medium is required.
Communication media can be cables carrying electrical signals, the atmosphere (or space vacuum) where radio waves circulate, or optical fibers that propagate light waves.
The Public Switched Telephone Network (PSTN) was the first medium used, because it existed even before computers. On the subscriber’s side, the network ends with a pair of copper wires connected to a switch. The telephone connected to it converts the speech signal into an electrical signal. The signal then reaches the switch, which directs it to another subscriber, possibly through other switches. France is planning to shut down its PSTN-type telephone network, but this does not mean the end of fixed telephone service: it will continue to be provided over next-generation networks (voice over IP), copper or fiber.
In order to connect computer equipment and transmit/receive digital data independently of conventional (i.e. analog) telephone services, a modem (modulator/demodulator) is used to convert the digital data from the equipment into a modulated signal that can be transmitted over an analog network, and vice versa.
The Integrated Services Digital Network (ISDN) is the digital equivalent of the analog telephone network. It uses the same physical infrastructure, but all signals remain in a digital form, making it more convenient for non-voice applications. It is therefore an extension of digital access to the subscriber.
Dedicated telephone lines, that is, telephone lines reserved for this purpose, have become necessary and popular since the 1970s. They make it possible to link several sites of a company, or the university campuses of a city, for example. They were also the basis for interconnecting large networks before fiber optics became essential.
The Power Line Carrier (PLC) has been used for some time, in low speed, for industrial applications and home automation (devices in the home are integrated into systems that need to communicate with each other in order to manage automation). The principle of PLC consists of superimposing a higher-frequency, low-energy signal on the 50 or 60 Hz alternating current. Therefore, existing electrical wiring is used.
The twisted pair consists of two insulated copper wires about 1 mm thick. These wires are helically wound, one on top of the other, to reduce the disturbing electromagnetic radiation found in parallel wires. The twisted pair can be used to transmit analog or digital signals and has a bandwidth of several Mbit/s over a few kilometers. Due to its satisfactory performance and low cost, the twisted pair is still widely used.
The coaxial cable is a cable with two conductors of opposite poles separated by an insulating material. The cable consists of a central conductor called the core, usually made of copper, which is embedded in a dielectric insulating material. The core is surrounded by a shield, which acts as a second conductor. In computer networks, the coaxial cable has been gradually replaced by fiber optics (for long-distance use, more than one kilometer) since the end of the 20th century.
An optical fiber is a very thin glass or plastic wire that can be a conductor of light and is used in data transmission. By convention, a pulse of light indicates a bit with a value of 1, and the absence of light, a bit with a value of 0. It is increasingly used by operators, in buildings, cities and even in underwater cables, to allow the interconnection of networks worldwide. Its throughput can reach 1 million Gb/s, which is its great advantage.
Cables have a major drawback: they are fixed and do not meet our mobility needs.
A wireless network is a network in which at least two devices can communicate without a wired connection. With wireless networks, a user has the ability to stay connected while traveling within a reasonably large geographical area.
Wireless technologies mainly use electromagnetic waves as a medium. The transmission and reception of these waves is carried out by antennas, integrated in wireless cards. The waves have a defect: they attenuate with the distance they travel and with the obstacles (walls, etc.) they encounter. The use of radio waves for data transmission is becoming increasingly widespread: cell phones, satellite communications, connected objects, etc.
There are several wireless network technologies, differing in the frequency of transmission used and the speed and range of transmissions. The three main standards can be selected according to the geographical area offering connectivity (coverage area): Bluetooth, Wi-Fi and GSM.
Bluetooth is a communication standard that allows the bidirectional exchange of data over very short distances. It has a theoretical data rate of up to 2 Mb/s and a range of 50 m to 100 m. It is often present on devices that operate on battery power and wish to exchange a small amount of data over a short distance: cell phones, laptops and various peripherals (mouse, keyboard, etc.). We are therefore in the field of wireless personal area networks (WPAN).
Wi-Fi is a set of wireless communication protocols governed by the IEEE 802.11 group of standards. With Wi-Fi standards, it is possible to create high-speed wireless local area networks. The range can reach several dozen meters indoors (generally between 20 and 50 meters), if there are no obstructions (concrete walls, for example) between the transmitter and the user. In practice, Wi-Fi makes it possible to connect laptops, office machines, personal digital assistants (PDAs), communicating objects or even peripheral devices, with a very high-speed connection. It is the protocol we use most often in our homes.
GSM (Global System for Mobile Communications) is a digital standard for mobile telephony. The third generation of mobile telephony (3G), whose main standard is UMTS, has made it possible to significantly increase the available bandwidth. Finally, 4G technology is the new generation, which is expanding around the world, with a throughput of up to 1 GB/sec. 5G technology, which is in preparation, will make it possible to download a film in a few seconds and will open up the market to new applications. We are in the field of wireless wide area networks (WWAN).
Even if the main activity of telecommunication satellites is the broadcasting of television programs, they are also used for mobile applications, such as communications to ships or airplanes. However, this could soon change, with operators in many countries launching into the race for mega satellite constellations providing Internet coverage to the entire world from space.
New technologies (e.g. the Loon project launched by Google, which involves deploying Internet coverage to areas that are very difficult to access, via balloons floating at an altitude of 20 kilometers), as well as new uses are being prepared.
The network card is the most important component. It is indispensable: all the data that needs to be sent and received from a network in a computer passes through it. The MAC address (Media Access Control), composed of 12 hexadecimal characters1, is the physical address of the card, a unique and worldwide address assigned at its manufacture. Your personal computer, smartphone and Wi-Fi box have a MAC address that should not be confused with the address relating to the network (e.g. IP address, which we will discuss in section 2.4.5).
A repeater is an electronic device combining a receiver and a transmitter, which compensates for the transmission losses of a medium (line, fiber or radio) by amplifying and possibly processing the signal, without modifying its content. It is used to duplicate and readapt a digital signal, to extend the maximum distance between two nodes in a network.
A hub (or concentrator) is a piece of hardware that concentrates network traffic from multiple hosts. It has as many ports to connect machines to each other (usually 4, 8, 16 or 32) and acts as a multi-socket to broadcast the information it receives from one port, to all of the other ports. Thus, all the machines connected to the concentrator can communicate with each other.
If the hub is unable to filter the information and transmit it to all the machines connected to it, the switch only directs the data to the destination machine based on its address. If computer 1 sends data to computer 2, only computer 2 will receive it.
A router is a piece of computer network interconnection equipment used to route data between two or more networks, in order to determine the path that a data packet will take. It is used to connect two different networks. For example, it is the boundary between the local network and the external network (Internet or other).
The topology of a network corresponds to its physical architecture. We can retain the following main topologies.
A bus topology is the simplest organization of a network. In a bus topology, all computers share a single transmission line (the bus) via a cable, usually coaxial. This is the common topology of an Ethernet-type local area network.
It has the advantage of being easy to implement and has a simple function. On the other hand, it is very vulnerable because if one of the connections is faulty, the whole network is affected. In addition, the transmission speed is low because the cable is common.
In a star topology, the computers in the network are connected to a central hub or switch system. Networks with a star topology are much less vulnerable because one of the connections can be disconnected without crippling the rest of the network. However, communication becomes impossible if the central element is no longer working.
In a ring network, all entities are connected together in a closed loop. Data flows in a single direction, from one entity to the next. At any given moment, only one node can transmit on the network and there can be no collision between two messages, unlike the bus-type network. This topology is used by the Token Ring and FDDI networks.
In a hierarchical topology, also called a tree topology, the network is divided into levels. The top is connected to several nodes lower down in the hierarchy. These nodes can themselves be connected to several nodes below them. The weak point of this type of topology is the “parent” computer in the hierarchy, which, if it fails, paralyzes part of the network.
In a meshed topology, each terminal is connected to all of the others. The disadvantage is that the number of connections required becomes very high. Indeed, the number of cables is n (n - 1)/2, if n is the number of computers. For example, it takes 28 cables to interconnect 8 computers, so this topology is used very little.
Hybrid topologies, combining several different topologies, are the most common. The Internet is an example.
The presence of a multitude of terminal equipment makes it necessary to define a coherent identification system within the network to differentiate them; this is called addressing. In addition, the network must be able to route information to any addressee according to their address: this is the routing function. When you put a letter in a mailbox, with the recipient’s address, this letter will be picked up by an employee of the company in charge of its routing (e.g. the postal service) and transported to a sorting center. Routing operations, sometimes complex, will allow this letter to arrive at the sorting center, in which the recipient is identified by their postal code. Therefore, a destination address and routing system are required. If you add your address on the back of the envelope, the addressee will be able to reply in the same way.
Early computer networks shared the same protocol and namespace. Each computer had a name, and all of the names were collected in tables that were installed on all members of these networks, which allowed routing. In order to communicate with another network, a computer that was a member of both networks had to act as a gateway and translate addresses from one to the other. It was fairly simple because there were only a few thousand computers at most. This is what I experienced in the late 1980s with the interconnection of the IBM and Digital Equipment “worlds” in universities.
The arrival of the Internet and the passage to millions of interconnected devices complicated the situation. Very precise addressing rules were developed little by little. We will discuss this further in section 2.4.4.
Switching is necessary when a call is made over several links in succession. Intermediate equipment associates an (inbound) link with another (outbound) link among those available.
In circuit switching (analog process), all of the links (the circuit) used for one communication are reserved for that one communication for its entire duration. Its concept and implementation simplicity made it successful in its use in the first communication networks, such as the telephone. It was the responsibility of the operators of a telephone switchboard to establish communications between users in the early decades of the telephone.
In message switching (digital process), there is no reservation of resources. A connection is only used by a communication during the periods of transmission of these messages. Messages from other communications can use the same links during this communication. Messages arriving at the switching node are processed in the order of arrival, which can generate queues.
Packet switching (digital process) uses the same principle as message switching, but the messages are made up of a succession of packets, whose size is better suited to the efficiency of the transmission. It is the most commonly used process in networks like the Internet. The problem that needs to be solved is the reassembly of the packets that make up the message.
There are two types of network architecture: client–server and peer-to-peer.
In client–server architecture, client machines (machines that are part of the network, such as a personal computer) communicate with a server (usually a powerful machine) that provides services (such as access to files, or a mail server). When the server has responded to the client’s request, the connection is terminated. There are countless examples of these communications, such as consulting a train schedule on the railway company server from your personal computer, or your phone. The client/server model has become one of the main concepts in network architectures.
The term peer-to-peer is often abbreviated to P2P. In a peer-to-peer system, nodes are simultaneously clients and servers of other nodes on the network, unlike client–server systems. The particularity of peer-to-peer architecture is that data can be transferred directly between two stations connected to the network, without passing through a central server. Peer-to-peer systems therefore facilitate the sharing of information and can be used, for example, for file sharing or distributed computing.
Imperfections in telephone conversations are usually not a problem, but this is not the case for data transmission, as the data must arrive at its destination complete and intact. The equipment involved (transmitter, receiver) must ensure this. The quality of service (QoS) of a data circuit is measured using several criteria:
The quality of service is subject to precise technical measurements, but for a user, it is quite subjective because it depends on their expectations and the type of network usage they have at any given time. For example, response time may seem acceptable if the user is looking at bus schedules, but completely unacceptable if they are participating in a videoconference, because it can greatly disrupt the flow of exchanges.
If an operator tells the users that the fiber optic connection in their home provides a speed of several hundred million bits per second, has the response time, in their use of the network, improved significantly compared to the connection they had previously? Not necessarily, because any network transaction involves a lot of intermediate equipment and many network sections with different speeds and congestion rates. The “effective” throughput will therefore depend on many parameters and may vary depending on the period and type of transaction.
Communication media “physically” connect equipment. As in any communication, a method is needed so that two entities can understand each other. A communication protocol is a set of rules that define how communication between two entities in a network should take place. Some of the important functions of a protocol include:
The protocols are hierarchically layered, with each one having to deal with specific functions.
In the 1960s, computing was centralized; that is, data was managed on “mainframes” that could be accessed by remote stations. These computers were linked together by networks operating on the basis of protocols developed by their manufacturers. The two most important protocols of this type are DECnet and SNA.
DECnet is a layered network architecture, based on a protocol defined by the Digital Equipment Corporation, the first version of which, in 1975, allowed two PDP-11 minicomputers to communicate. Large networks of computers from this manufacturer, particularly VAX machines, were deployed until the arrival of TCP/IP protocols.
SNA (Systems Network Architecture) is a layered network architecture defined by IBM in 1974. It is a functional architecture of the same type as the OSI reference model (which it precedes by seven years) and is also part of the IBM product family. Like DECnet, it is a proprietary architecture. SNA has been widely used by computer centers in banks, financial institutions and research centers equipped with IBM hardware.
The major flaw of proprietary architectures, such as those that will come to be cited, is that it is not easy to make them communicate with each other, unless an agreement is reached and a communication protocol is written between these architectures.
To solve this problem, in the 1970s, the ISO (International Organization for Standardization) developed a reference model called the OSI (Open Systems Interconnection). This model described the concepts used to standardize the interconnection of systems. It was organized in seven distinct layers, each bearing a number, ranging from the most abstract data (layer number seven) to physical data (layer number one). The OSI standard was published in 1984.
Let us quickly describe the seven layers:
In the United States, the Defense Advanced Research Projects Agency (DARPA), which is responsible for military defense research projects, launched a computer network project in 1966 linking certain American universities. In 1980, this network, called ARPAnet, became a military issue and was divided into two: the university network became NSFnet, funded by the NSF (National Science Foundation). ARPAnet became the heart of the future Internet and a tool for the development of this new technology.
The NSFnet network opened up to the world, and interconnection problems soon emerged. Communication between networks using different architectures (proprietary or not) became too complex.
Extensive research and development work in 1977, in which the differences between the protocols were blurred by the use of a common communication protocol, led to the demonstration of a prototype, called TCP/IP. On January 1, 1983, TCP/IP officially became the only protocol on ARPAnet. The Internet (from “inter-network”), takes the meaning of a worldwide network using the TCP/IP protocol.
The Internet is not a new type of physical network. It offers, through the interconnection of multiple networks, a global virtual network service based on TCP (Transmission Control Protocol) and IP (Internet Protocol) protocols. This virtual network is based on a global addressing that is placed above the different networks used. The various networks are interconnected by routers. Thanks to the growing interest in vast communication networks and the arrival of new applications, Internet techniques have spread to the rest of the world.
The OSI model has been developed with a normative vocation (i.e. to serve as a reference in the course of communication between two hosts), whereas the TCP/IP model has a descriptive vocation (i.e. it describes the way in which communication takes place between two hosts).
TCP/IP actually refers to two closely related protocols: a transmission protocol, TCP (Transmission Control Protocol), which is used over a network protocol, IP (Internet Protocol). It is also a set of protocols that are generally used at the application layer, using TCP/IP.
The TCP/IP model is simpler than the OSI model, with only four layers:
1) the network layer includes the physical and data link layers of the OSI model. The only constraint of this layer is to allow a host to send IP packets over the network;
2) the Internet layer is the cornerstone of the architecture. Its role is to allow the injection of packets into any network and the routing of these packets, independently of each other, to their destination. The Internet layer has an official implementation: the IP protocol;
3) the transport layer has the same role as that of the OSI model: to allow even entities to support a conversation. This layer has a main official implementation: the TCP protocol, a reliable, connection-oriented protocol that allows the error-free routing of packets from one machine on the Internet to another machine on the same Internet;
4) the application layer contains all high-level protocols, such as Telnet, SMTP (Simple Mail Transfer Protocol) and HTTP (HyperText Transfer Protocol). It has indeed been noted with use that network software rarely uses the presentation and session layers of the OSI model.
TCP/IP is an open protocol that is independent of any particular architecture or operating system. This protocol is also independent of the physical medium of the network. This allows TCP/IP to be carried by different media and technologies.
Each piece of equipment on a network is identified by an address, called the IP address. The addressing mode is common to all TCP/IP users regardless of the platform they use.
The MAC address, already mentioned, is a unique identifier assigned to each network card, but in a large network, there is no central element that knows the location of the recipient and can send the data accordingly. The IP address system, on the other hand, is used in a process called routing to ensure that the data reaches the recipient. Currently, two versions of IP coexist: IPv4 and IPv6.
In IPv4, the addresses are exactly 32 bits (4 bytes): enough to code 4,294,967,296 different IP addresses. The IP address is composed of four groups of decimal digits, noted from 0 to 255, and separated by a dot (e.g. 220.127.116.11). It is the most widely used protocol in the world. It is used for both local IP addresses and public IP addresses.
To identify each other, the computers that make up the Internet network essentially use a series of numbers, each number (IP address) corresponding to a separate machine. The Internet Corporation for Assigned Names and Numbers (ICANN) coordinates these unique identifiers internationally and brings together, in a non-profit partnership, people from around the world who work to maintain the security, stability and interoperability of the Internet.
Often, in order to connect to a computer server, the user does not give his IP address, but his domain name. A domain is a set of computers connected to the Internet with a common characteristic. The domain name system is hierarchical, allowing the definition of sub-domains whose codes (levels) are separated by a dot. For example, the domain inp.cnrs.fr designates the CNRS Institute of Physics in France. The rightmost part, such as “com”, “net”, “org” and “fr”, is called the top-level domain. The domain name is then resolved to an IP address by the user’s computer using the Domain Name System (DNS). It is possible to initiate a connection only once the address is obtained.
As the structure of the IPv4 address no longer made it possible to respond to all address requests, it was necessary to develop a new structure called IPv6, with 128 bits. This makes it possible to have over 256 billion billion billion billion billion different IP addresses!
The growing importance of the Internet has led to its very precise organization. The three main regulatory bodies are as follows:
The growth of the Internet has been extraordinary, with 10,000 computers connected in 1987, 2.5 million in 1994, 17 million in 1997, 400 million in 2000 and 3.5 billion in 20173. This development concerns the entire planet, as shown in Figure 2.14.
For France, in 2017 the ARCEP (Autorité de Régulation des Communications Electroniques et des Postes) announced 25 million people connected to high (ADSL)- or very high (fiber optic)-speed Internet, and about 70 million SIM cards providing access to the Internet.
E-mail and file transfer are the oldest applications on the Internet. But the service that made the Internet popular with the general public is the World Wide Web, which began to spread in 1993 (more details on this are given in section 2.5.1). The rapid increase in the capabilities of computers meant that they were capable of encoding and processing sound or voice, as well as still images or video.
But interactive multimedia applications, such as videoconferencing, need efficient group (multi-user) transmission on the one hand, and performance guarantees on the other hand. Since the Internet is a network that provides a routing service without any guarantee of performance (Best Effort Principle), it was necessary to develop control mechanisms that allowed multimedia applications to adapt their behavior according to the conditions of the network.
The evolution of the Internet is taking place in parallel with an explosion of new applications, which seek to make the best use of the services available at any given moment. Examples include games distributed over the Internet (in which several thousand players around the world can compete on a battlefield or in a board game), or collaboration tools (distance learning, collaboration of doctors around scanner/X-ray images visible and writable by all). More generally, we can expect the Internet to represent a revolution that is at least comparable to the telephone revolution that began in the last century.
We are seeing that computer networks, especially the Internet, are having an ever-increasing impact on our daily lives. For better or for worse? In 2017, the Internet Society released a report entitled “Paths to our Digital Future”5.
This report analyzes the key driving forces that will have a profound impact on the future of the Internet in the near future:
The report analyzes three areas of impact:
The above tells the story that led to the standardization of communications between computers of all sizes. But networks have changed profoundly: the volume of data traffic, the very rapid increase in the number of sites, broadband (20 Mbit/s at home), transporting multimedia data on the same medium (telephone, television, games, information, etc.), wireless mobile access, etc.
Here are some major areas of application, of which we will see more specific examples in Chapter 6.
The Web was invented in 1989 at CERN (European Organization for Nuclear Research), based in Geneva, by a British physicist, Tim Berners-Lee6. Originally, the project, called the World Wide Web or W3, was designed and developed with his colleague Robert Cailliau so that scientists working in universities and institutes around the world could exchange information instantaneously. On April 30, 1993, CERN put the World Wide Web software in the public domain. Tim Berners-Lee left CERN to go to the Massachusetts Institute of Technology (MIT) in 1994, where he founded the World Wide Web Consortium (W3C), an international community dedicated to the development of open web standards.
We all use the Web, without necessarily knowing it. A website is nothing more or less than a collection of files stored on a web server. Web browsers are applications that retrieve the content of pages, located on web servers, to send them to another computer, the latter being called a web client.
The Web is based on three main ideas: hypertext navigation, multimedia support and integration of pre-existing services (e-mail, file transfer, etc.). When writing a document (called a page) on the Web, certain words can be identified as access keys and a pointer to other documents can be associated with them. These other documents can be hosted on computers on the other side of the world.
In October 1990, Tim Berners-Lee described the three technologies that remain the foundation of today’s Web:
HTML was invented to allow writing hypertextual documents, also called web pages, linking the different resources of the Internet with hyperlinks. It is a so-called markup language (or structuring language), whose role is to formalize the writing of a document with formatting tags. The tags make it possible to indicate the way in which the document should be presented and the links it establishes with other documents.
Here is an example of an HTML file, with a title and a body of two paragraphs, one of which contains a hyperlink:
<!DOCTYPE html> <html> <head. <title>Example HTML file</title> </head> <body A sentence with a <a href=“target.html”>hyperlink</a>. <p> A paragraph where there is no hyperlink. </p> </body> </html>
Since HTML does not attach to the final rendering of the document, the same HTML document can be viewed using a wide variety of hardware (computer, tablet, smartphone, etc.), which must have the appropriate software (web browser, for example) to provide the final rendering.
A web browser is a software program designed to view and display especially HTML pages. Technically, it is at least an HTTP client. Let us mention a few web browsers: Netscape (1994), Internet Explorer (1995), Mozilla (1998) and Firefox (2005), Safari (2003), etc.
The standards for this language have evolved (successive versions) to take into account the new possibilities offered by Internet navigation.
HTTP (Hypertext Transfer Protocol) is a communication protocol developed for the World Wide Web. It was designed for the transfer of hypermedia documents such as HTML. It follows the classic “client–server” model, with a client that opens a connection to send a request, then waits until a response is received. HTTPS (with an S for secured) is the secure HTTP variant.
The best known HTTP clients are web browsers. The user’s computer uses the browser to send a request to a web server. This request asks for a document (e.g. an HTML page, an image, a file). The server looks for the information to finally send the response.
Website addresses, also called URL (Uniform Resource Locator) addresses, look more or less like this: http://www.example.com. Every document, image or web page has a URL address, which is often used to link to it.
A URL consists of at least the following parts:
We use less and less directly URLs due to our intensive use of search engines such as Google or Yahoo.
The electrical equipment in our homes uses energy that comes from “somewhere”. We do not have to worry about the source of this energy, which is diversified and can change depending on the period, since these sources (nuclear power plants, hydroelectric power plants, etc.) are interconnected by networks that guarantee that we are always supplied.
Cloud computing uses the metaphor of clouds to symbolize the dematerialization of computing. It involves moving IT services to remote servers, managed by suppliers and accessible via the Internet, and thus having access to virtually infinite services and resources (storage, computing).
Cloud computing is of interest to individuals (e.g. to store photos and videos), small businesses (which thus have access to resources they could not afford) and large companies alike.
There are three types of cloud computing:
The client cannot locate the physical sites that host these services, and these sites are subject to change. The advantages are numerous: cost reduction by shifting IT to the provider, ease of use from any location thanks to the Internet, quality of service, flexibility to manage peak loads, etc. Application and data security is, of course, a critical aspect and it is essential to address it. Some companies avoid this solution for storing and processing highly sensitive data.
The cloud computing market is huge, and there are many solution providers: major computer manufacturers (IBM, HP, etc.), Amazon, Google, Microsoft, OVH in France to name only the most significant. According to Synergy Group, the turnover of cloud computing suppliers reached 180 billion dollars for the period October 2016–September 2017 with an overall growth of 24%.
While the Internet was designed for humans to communicate and access information, the idea is that objects can exchange information and humans can acquire information through objects.
Imagine a world where all objects are able to exchange information and communicate with each other, as well as to communicate and interact with their users through the Internet and other less well-known but equally effective communication networks. This is the world of the Internet of Things.
A connected object has the ability to capture data and send it, via the Internet or other technologies, for analysis and use. These billions of connected objects will create an exponential volume of data that will need to be stored, analyzed, secured and restored for various uses.
Analysts predict more than 50 billion connected objects in a few years. This connected and intelligent world is expected to explode the volume of data from 8 zettabytes (8 trillion billion) in 2015 to 180 zettabytes in 2025, 95% of which will be unstructured data (text data, JPEG images, MP3 audio files, etc.); a volume that is expected to be 92% processed in the Cloud (Huawei’s prospective report).
In terms of technologies, standardized wireless access (such as Wi-Fi and Bluetooth) currently dominates due to the strong development of consumer applications (connected home, sports/wellness, electronic gadgets) and of course its low cost. New technologies and protocols have been and are still being developed to take into account the constraints and specificities of the many areas of IoT use (energy consumption, for example). We mention below the most used ones.
NFC (Near-Field Communication) is a technology that allows data to be exchanged at a distance of less than 10 cm between two devices equipped with this device. NFC is integrated in most of our mobile terminals in the form of a chip, as well as on certain transport, payment or access control cards for restricted access premises. The reader can simply operate the unlock or be connected to a network to transmit the information corresponding to your entry. In the latter case, you enter the IoT domain.
RFID (Radio Frequency Identification) is a technology that enables memorizing and retrieving data remotely using radio tags (RFID tags) that can be stuck or embedded in objects and even implanted in living organisms (animals, human body). The reading of passive chips can extend up to 200 meters. This technology is widely used in business.
Low-Energy Bluetooth (also known by the acronym BLE and Bluetooth Smart) is replacing NFC and is mainly intended for connected objects where the need for throughput is low and battery life is crucial, as well as nomadic equipment such as smartphones, tablets, watches, etc. The range can be counted in a few dozen meters.
Short-range radio protocols (ZigBee, Z-Wave) are intended for the creation of private local area networks (for home automation, for example). They are energy efficient and offer high data rates.
Low-speed radio protocols (Sigfox, LoRa) are particularly suitable for energy-efficient equipment that emits only periodically, such as sensors.
LTE-A, or LTE Advanced, is a fourth-generation cell phone network standard. It gives the IoT much more performance, and its most important applications concern vehicles and other terminals in motion.
The value added by the Internet of Things is in the new uses it will bring. Let us retain the most important ones; we will detail some of them in Chapter 6.
The sector is carried by the smart home: connected security devices (wireless surveillance cameras, alarms, etc.) and those dedicated to the automation of the home (thermostats, locks, intercoms), not forgetting the large connected household appliances (refrigerators, washing machines, etc.) and robots. The “smart city” is another important area: road traffic, transportation, waste collection, various mapping (noise, energy, etc.).
Wearable technologies are developing: connected watches and glasses, smart clothing, etc. They are also found in the monitoring of our health (connected scales, monitoring of patients with chronic diseases), in leisure and sports and in many toys.
Other areas include: environmental monitoring (earthquake prediction, fire detection, air quality, etc.), industry (measurement, prognosis and prediction of breakdowns), logistics (automated warehouses), vehicles that require more autonomy and robots in various environments.
Here is now a very common example: contactless means of payment using a bank card, a cell phone or a bracelet which communicates with payment terminals using the NFC communication protocol already mentioned; we therefore avoid inserting the bank card and entering a confidential code. This protocol allows data to be exchanged at a very short distance (a few centimeters). A chip and an antenna are integrated into your bank card, your cell phone, etc. Via a smartphone, the applications dedicated to NFC payment can also include a certain number of additional and very useful functions such as the automatic taking into account of loyalty cards in stores. But do we know all the features and what is done with the information collected?
Each object has an often simple function. But if several objects can be made to collaborate, to make them interoperable, their capabilities will be considerably increased. This is, for example, the ability of industrial robots to communicate directly with each other or the ability of different connected objects involved in flow management (factory, hospital, etc.).
A major obstacle: the connectivity of objects is dominated by proprietary technologies, often developed without technical or legal standardization.
Whether in the medical field (patient tracking devices), the automotive industry (connected cars), agriculture (precision farming) or home automation, devices that take advantage of the Internet of Things generate an unprecedented amount of data. These data are often confidential and personal. Are they really protected?
When you turn on your connected speaker, what is said in the room is recorded somewhere. Your question, “What will the weather be like tomorrow?” will be analyzed by the computer system to understand it and provide you with the answer. But your conversation will also be recorded if you are not careful. Who can use this information and for what purpose?
Moreover, connected objects represent a risk in terms of cybersecurity. These devices are designed to be as simple as possible, in order to limit their cost and facilitate their use. However, this simplicity also makes them more vulnerable than other electronic devices such as smartphones.
Gartner Inc., a US-based advanced technology consulting and research firm, announced that global spending on Internet of Things security was expected to reach $1.5 billion in 2018.
The multiplication of connected objects leads to an important aspect of the development of computing. We have gone from “mainframe computers” in the hands of specialists (computer scientists) to personal computers that can be used simply by anyone thanks to highly efficient graphical interfaces. We are entering a third era, one in which computers are disappearing, leaving us in a hyper-connected world in which the computer is invisible.
This vision of ubiquitous computing (the term is derived from the Latin ubique meaning “everywhere”), which is constantly available, was first formulated in 1988 by Mark Weiser of the Xerox Palo Alto Research Center. It is also referred to as ambient intelligence and pervasive computing.
In Mark Weiser’s idea, computer tools are embedded in everyday objects. The objects are used both at work and at home. According to him, “the deepest technologies are those that have become invisible. Those which, knotted together, form the fabric of our daily life to the point of becoming inseparable from it”.
Today’s IT systems are decentralized, diverse, highly connected, easy to use and often invisible. A whole range of discrete devices communicate discreetly through a fabric of heterogeneous networks.
Spontaneous and autonomous networks, also called ad hoc networks, have an important place in this development. Ad hoc networks (Latin for “who goes where he must go”, i.e. “formed for a specific purpose”, such as an ad hoc commission formed to solve a particular problem) are wireless networks capable of organizing themselves without a predefined infrastructure.
The first research on “ad hoc multi-hop” networks dates back to the 1960s and was carried out by DARPA, as the military was very interested in this approach for the battlefield.
Ad hoc networks, in their most common mobile configuration, are known as MANET (for Mobile Ad hoc NETworks). MANET is also the name of an IETF working group, created in 1998–1999, tasked with standardizing IP-based routing protocols for wireless ad hoc networks.
A MANET network is characterized by:
These technologies are particularly used in sensor networks. The sensor is a device that transforms the state of an observed physical and/or logical quantity into a usable quantity. Wireless sensor networks are considered a special type of ad hoc networks where fixed communication infrastructure and centralized administration are absent and nodes play the role of both hosts and routers.
This type of network consists of a set of micro-sensors scattered across a geographical area called the catchment field, which defines the terrain of interest for the phenomenon being captured. The deployed micro-sensors are capable of continuously monitoring a wide variety of ambient conditions.
Sensor networks respond to the emergence of an increased need for diffuse and automatic observation and monitoring of complex physical and biological phenomena in various fields: industrial (quality control of a manufacturing chain), environmental (monitoring of pollutants, seismic risk, etc.), security (risk of failure of large-scale equipment such as a dam), etc.
Networks are fuelling ever-increasing cybercrime, not to mention their use for malicious political purposes, which has been in the news since 2017.
In May 2017, hackers attacked thousands of governments and businesses around the world with malware, blocking the use of computers and demanding ransom. Considering the largest ransomware cyber attack in history to date, WannaCry infected more than 300,000 computers across more than 150 countries in a matter of hours.
The US-based credit company Equifax was the target of a massive hacking attack in 2017. The information of more than 140 million Americans and more than 200,000 consumer credit card numbers were accessed by hackers. This attack exploited a vulnerability in one of the company’s applications, allowing access to certain secret files.
The main objectives of computer attacks are:
There are several types of malware:
A denial of service attack is a type of attack designed to make an organization’s services or resources (usually servers) unavailable for an indefinite period of time. There are two types: denial of service by saturation and denial of service by exploiting vulnerabilities.
It is essential to put in place measures to secure networks to ensure:
We have the ability to protect our personal network or that of our organization against certain attacks from outside. Several complementary methods are available. We can limit communication and visibility from the outside. The most commonly used method is the implementation of a firewall that forces the passage through a single point of control (inbound and outbound: Who? For what?). Its main task is to control traffic between different trusted zones by filtering the data flows that pass through them.
Encryption is a process of cryptography through which we wish to make the understanding of a document impossible for anyone who does not have the decryption key. Cryptography is very old and children, even today, can still have fun encrypting messages, with simple codes, so that their parents do not understand the meaning!
There are two main families of encryption: symmetric and asymmetric encryption. Symmetric encryption allows us to encrypt and decrypt content with the same key, known as the secret key. Symmetric encryption is particularly fast, but requires the sender and receiver to agree on a common secret key or to transmit it via another channel.
Asymmetric encryption assumes that the (future) recipient has a pair of keys (private key, public key) and has ensured that potential senders have access to its public key. In this case, the sender uses the recipient’s public key to encrypt the message while the recipient uses his private key to decrypt it.
It is essential for everyone to ensure the security of their IT environment, even if it is just a personal computer. Here are a few simple rules: