The Internet of Things and Ubiquitous Intelligence (1)

Editor's Desk:
The traditional Internet is oriented towards person-to-person connection, whereas the Internet of Things is oriented towards connection of inanimate objects. As such, the Internet of Things covers a larger range of connections and involves more semantics. Internet and telecom networks are focused on information transfer, while the Internet of Things is focused on information services. By combining sensor networks, the Internet, telecom networks, and cloud computing platforms, the Internet of Things can sense, recognize, affect, and control the physical world. The physical world can be unified with the virtual world and human perception. This lecture discusses Internet of Things technology from three perspectives: Ubiquitous information sensing, ubiquitous network convergence, and intelligent information services. In this part, we will introduce the technical characteristics of the Internet of Things and sensor networks, the development background of sensor networks, and key technologies of sensor networks.

1 Introduction

1.1 Technical Characteristics of the Internet of Things and Sensor Networks
    Inspired by new technologies in computing, microelectronics, communications networking, and man-machine interaction, society has shifted its focus from network connection services to ubiquitous information services. With this shift, the Internet of Things has developed rapidly. However, as its technical connotations and denotations are continuously evolving, there is currently no uniform, integral, or accurate definition. People are defining the Internet of Things differently during its different stages of development. On  November 17, 2005, the ITU released ITU Internet Report 2005: The Internet of Things [1], where the Internet of Things was defined as Radio Frequency Identification (RFID)-based thing-to-thing and thing-to-person Internet. But as sensor networks have matured and modern service technologies and new services have emerged, traditional definitions of the Internet of Things no longer satisfy technical and application requirements.

    The Internet of Things is generally thought to be a network integrating information collection (using wireless sensor networks), information transmission (using the Internet and telecom, radio, and TV networks), new services and applications based on information service networks are also part of the Internet of Things. Generally speaking, the Internet of Things has typical characteristics:

    (1) Ubiquitous information capturing. The sensor network is the engine of the Internet of Things and is at the tip of information sensing and processing. What makes sensor networks distinct from other networks is their capacity to collect massive amounts of diverse information on environment and events in close proximity. This information is obtained from various perspectives and for multiple parameters. The diversity of information, mass quantity of information, and complexity of relationships between information is unprecedented. With node identification, sensor devices can assign specific locations and identifications to a piece of information, which allows the information to be used in future applications.

    (2) Reliable information transmission. In the Internet of Things, information categories become much richer and differentiation in Quality of Service (QoS) becomes more complicated than in existing networks. Information service rather than connection service will be a basic operation feature of the Internet of Things. As an infrastructure and support environment for a ubiquitous information society, ubiquitous networks will be an important objective in developing information communication networks. Reliable and effective transmission technologies are required in existing networks for providing ubiquitous and intelligent services and providing people with rich real-world information.

    (3) Efficient information applications. The major difference between the Internet of Things and existing networks is ubiquitous intelligence. The richness that comes with masses of sensing information and reliable information transmission is essential for providing diverse services using the Internet of Things. Processing and utilizing mass sensing information can provide enterprises and the general public with new service modes and experiences—especially ubiquitous services, business operation modes, and management systems that can adapt to context.

    This lecture discusses three important components of the Internet of Things: Sensor network collection of information, ubiquitous network-based information transmission, and new service technologies and businesses using collected information. This part focuses on the sensor network.

1.2 Development Background of the Sensor Network
    Since the middle of the 20th century, computing technologies and network technologies have together given rise to many new application modes and technical means. But all these fall into the category of traditional person-to-person or person-to-machine interaction. With the rapid development of communication technologies, embedded computing technologies, and microelectronic technologies, the manufacturing costs of electronic products has been steadily decreasing, as predicted by Moore’s Law. In the 1990s, the first microintelligent sensor was put into use that had sensing, computing, and communication capabilities and was suitable for different environments. Using the microsensor as the network node—Wireless Sensor Network (WSN) technology—completely changed the way information is acquired [2].

    WSN is a potential network. Many simple nodes can be randomly deployed in remote locations and under adverse conditions. In a self-organizing network, all nodes work cooperatively to monitor, sense, and collect information in complicated environments or to monitor objects within the network’s distribution area and process the information. In this way, nodes obtain detailed and accurate information and then send it to interested observers. Because it is a bridge between the objective physical world and subjective sensing world, WSN is a new information acquisition and processing technology, a revolution in information sensing and collection.

    Now, WSN is a hot research topic in network communications and information processing in the U.S., and efforts to develop it are being stepped up. In its Information and Communication Technologies (ICT) Seventh Framework Programme (FP7) started in 2007, the European Union lists WSN as an important research area in network embedded control systems. In South Korea, the IT839 strategy makes ubiquitous sensor networking one of the three key infrastructures. In Japan, both e-JAPAN and U-JAPAN strategies have sensor networking as a goal for creating a next-generation Information and Communication Technology (ICT) society by 2010.

2 Key Technologies of the Sensor Networks

2.1 Nodes
    Nodes are the basis for WSN research and application. All protocols, mechanisms, and algorithms related to WSN must be implemented via nodes. Therefore, node design and implementation determine functions, performance, and investment across the entire network. Most WSNs use general embedded platforms as their nodes; for example, Crossbow’s Mica-and Telos-series nodes. These platforms use common CPUs to connect peripheral devices—such as sensors and RF chips—in order to form sensor nodes. Node functions are developed mainly by software. On the one hand, node functions, supported by TinyOS and other operating systems, are relatively easy to develop. On the other, node development is limited because of fixed hardware and node size, and cost. Also, power consumption cannot be reduced through physical design. Nodes have limited processing capability and it is difficult to develop sophisticated programs to counter this.

    Most WSN applications rely on critical performance indexes for achieving low power consumption, low cost, small size, easy deployment, and reliability. One practical approach to meeting these demands is to employ System on Chip (SoC) nodes. Highly integrated technology and good physical design can reduce size, cost, and power consumption of nodes, and architecture designed especially for WSN improves its computing efficiency and capability. This is a trend for future node technologies.

2.2 Network Protocol and Networking
    In WSN, the protocol stack design determines the self-organization mode and communication performance of network nodes. It also determines interconnection and access methods of heterogeneous networks. Protocol stack design ensures effective energy, scalability and reliable transmission in WSN when there are limited network resources. WSN protocol stack and working technologies fall roughly into two categories: Non IP-based protocol stack and IP-based protocol stack.

    Two examples of non IP-based protocol stacks are Zigbee and Sensor-Net. Zigbee is a low consumption, low power, reliable wireless network standard from the Zigbee Alliance [3]. Sensor-Net is an architecture especially designed for the sensor network [4]. It adopts cross-layer design for energy management, system management, time synchronization, discovery, and security. Inter-layer cooperation allows better management and control.

    At present, there are only a few IP-based protocol stacks, including NanoStack, PhyNet, and IPv6 micro sensor router. NanoStack is an embedded sensor network software project based on IPv6 over Low power Wireless Personal Area Networks (6LoWPAN) [5]. This protocol stack contains User Datagram Protocol/Internet Control Message Protocol (UDP/ICMP), 6LoWPAN IPv6 protocol, and IEEE 802.15.4 Media Access Control (MAC) protocol. Because IPv6 allows for much larger packet capacity than the maximum frame capacity specified in IEEE 802.15.4, an adaptation layer is added between the network and data link layers to splice and reorganize IP packets. In this way, low-rate transmission requirements of Wireless Personal Area Networks (WPAN) can be met. PhyNet protocol stack is an industrial solution based on the IETF’s 6LoWPAN. It is used for implementing IP-based communication in WPAN. This protocol stack takes the 6LoWPAN standard as its core and supports IEEE 802.15.4 physical layer and MAC layer protocols.

    In addition, WSN must support interconnection with other networks for sharing and processing information. Further study into the architecture and self networking technologies of WSN is required.

2.3 Information Processing
    Information processing technologies of the sensor network are diverse, ranging from embedded, lightweight information processing technology within a node to distributed network information processing technology between nodes. On the one hand, WSN is cheap, has high observation precision, and can integrate heterogeneous sensor information. These advantages lay the foundation for implementing an ideal information processing technology. On the other hand, with low power consumption, heterogeneous interconnection, ubiquitous cooperation, and limited nodes, WSN has some technical drawbacks for information processing. The progress of research on core information processing technologies for the sensor network is as follows:

• Signal processing, information identification, and information extract technologies within a node: These technologies are used to guarantee the accuracy of signal collection and sensing. Especially for sensor networks with limited nodes, embedded and lightweight signal processing, information identification, and extract technologies are necessary. Node signal processing technology and information identification, information extraction, and application of technology are closely related. These technologies need to tune the node degree of intelligence and lightweight signal processing algorithms constrained by the node resource.
• Distributed cooperative processing technologies between the nodes: The goal of these technologies is to improve a node’s intelligence in estimating its surroundings and identifying attributes. Such technologies include cooperative technology among distributed nodes and multisensor information integration technology. With limited resources, a node in the sensor network can only sense local information. Hence, all nodes in the network must cooperate with each other to obtain global information. In a multisensor system, information is extremely diverse, the quantity of information is massive, and relationships between information are complex. Therefore, multisensor information integration technology is the key for information processing in WSN.
• Information compression and coding technologies. These can greatly improve the transmission performance of a multihop link and bring about effective, reliable signal transmission. Packet compression is an effective low consumption communication technology that reduces the length of a packet by information redundancy. It can shorten end-to-end transmission delay of a data packet and reduce conflict in accessing shared channels.

2.4 Security
    Due to resource limits, WSN has more security challenges than traditional networks. A low consumption security solution designed for WSN should take into account optimal loads for communication and computing, distribution and update of secure keys, security authentication and anticapture of nodes, and scalability in secure key management. The goal of WSN security technologies is to ensure information security during node communication.

    Key management is a basic but important step in security management. To improve security and antiattack performance of the network, a key management mechanism needs to be implemented. Current research into attacks against sensor networks focuses mainly on secure routing technologies. In China, researchers have been studying specific key management protocols—which are still theoretical models—rather than studying sensor network security from a systematic perspective. Moreover, research on secure routing technologies is concentrated on stability and simulation of routing algorithms. Although some security models and objectives have been proposed, there is still no secure routing approach that is feasible for sensor networks.

    In China, research into security models for sensor networks has been focused on particular scenarios rather than on a general security solution. Therefore, one area that requires systematic study is general security models for sensor networks, including secure communication protocols, secure key management, secure routing, and access control.

(To be continued)

References:
[1] International Telecommunication Union (2005) ITU Internet Report 2005: The Internet of Things. Geneva: ITU. [Online]. Available: http://www.itu.int/pub/S-POL-IR.IT-2005/e
[2] I. F. Akyildiz, W. Su, Y. Sankarasubramaniam, and E. Cayirci, “A survey on sensor networks,” IEEE Communications Magazine, no.8, Aug. 2002.
[3] ZigBee Alliance. [Online]. Available: http://www.zigbee.org/
[4] J. Polastre, J. Hui, P. Levis et al., “A unifying link abstraction for wireless sensor networks,” in Proc. 3rd ACM Conf. Embedded Networked Sensor System (Sensys05), San Diego, CA, 2005, pp. 
[5] “IPv6 over low power WPAN,” datatracker.ietf.org. [Online]. Available: http://datatracker.ietf.org/wg/6lowpan/charter/

Biographies

Dongliang Xie (xiedl@bupt.edu.cn) is a director and associate professor at the Broadband Network Center of State Key Laboratory of Networking and Switching Technology, Beijing University of Posts and Telecommunications. He researches wireless and mobile network technologies, wireless sensor networks, mobile Internet QoS, network cooperation, and ubiquitous intelligence. He has published more than 40 papers.

Yu Wang (wang0yu@gmail.com) is a master’s student at the State Key Laboratory of Networking and Switching Technology, Beijing University of Posts and Telecommunications. She researches convergence of wireless sensor network and ubiquitous network.