A sensor array is defined as a set of sensors connected to one another, each sensor being provided with a transceiver. Sensor networks are a new generation networks with specific properties, which do not fall within the conventional architectures.
The miniaturization of sensors poses problems of communication and power resources. It is necessary that the sensor is smart enough to gather the required information and the issue wisely. In addition, the sensor processor must not be used too extensively in order to consume the least amount of energy. It must therefore incorporate reactive elements rather than cognitive. Finally, to ensure a good rate, range transceivers must necessarily be low, on the order of ten meters. The establishment of a network of sensors thus arises routing problems, error monitoring and power management.
From a communication perspective, the IP protocol environment is too heavy and causes an excessive flow and consumption. The solutions that were derived from fieldbus or industrial real-time networks have a better compromise between efficiency and energy consumption. As the sensors can be distributed per hundred square meter, IPv6 addressing seems most likely. In the future, it will surely use an IP packet encapsulated environment in specific fields to determine.
For now, the problems of security and quality of service are put into the background compared to consumer problems. An important field of research is in any case open to make effective and resilient sensor networks.
The main concern ZigBee radio standard. Wibree and 6LoWPAN form other solutions. Wibree is a very low consumption technology with a range of 10 m and a flow rate of 1 Mbit / s. This solution was developed by Nokia to compete with both ZigBee and Bluetooth.
The 6LoWPAN networks (IPv6 over Low power Wireless Personal Area Networks) come from an IETF working group. The objective is clearly to allow continuity of IP towards low and powerful machines with power limited power.
The use of the IPv6 standard to obtain a very large number of addresses for Sensor Networks huge problem. Indeed, the sixteen bytes of address of the sender and the sixteen bytes of address of the receiver more required fields involve misuse of the radio link to transport these supervisory information. This becomes really problematic with little energy sensor. ZigBee, however, limits the length of his frame to 127 bytes, which can also cause problems if the information to be transported from a sensor is long.
These sensor networks form mesh networks, and they need a routing protocol. Using a protocol such as IEEE 802.11s associated with IPv6 addresses would be catastrophic for the battery life of the sensors. For this reason, the current proposals are much simpler, with protocols such as LOAD (6LoWPAN Ad hoc Routing Protocol), simplification of AODV, DYMO-Low (Dynamic MANET On-demand for 6LoWPAN), a simplification of Dymo MANET working group, and Hi-Low (Hierarchical Routing over 6LoWPAN), which has a hierarchical addressing. These different protocols from the IETF proposal and therefore the normalization of ad-hoc networks, but not taking into account all that is optional.
Another important feature of sensor networks protocols relates to the discovery service, which must allow the starting of the network automatically. The IETF also plays an important role in this area by proposing several solutions, one oriented sensors: LoWPAN Neighbor Discovery Extension. This protocol is a reduction of the standard Neighbor Discovery for all consumers of energy elements, such as broadcast and multicast management.
A particular sensor network is proposed by smart dust (Smart Dust), whose objective is to develop nanotechnology sensors and connect them by an ad-hoc or mesh-type network. Smart dust lies in a lower cube cubic millimeter, hence its name dust. In this dust, are all the components necessary to realize a computer communicating: a processor, memory, radio, battery, etc.
The main problem here is still saving energy when performing the functions of the sensor. In particular, the network portion must carry communications expending little energy. The University of Berkeley has designed an operating system named TinyOS specific protocols and Tiny protocol. The TinyOS has been written in a simplified C language called nesC, which is a kind of dialect designed to optimize memory usage.