Building wireless sensor networks

09/06/2009 by David Slade.

Competitive pressures, globalization, rising energy prices, and increasingly stringent regulations are driving companies to cut costs while increasing efficiency and productivity. Industry leaders feel pressure to adopt new technologies that will help them gain competitive advantages wherever they can find them. One of these promising new technologies is wireless sensor networking. Wireless sensor networks (WSNs) dramatically reduce the cost of installing and commissioning instrumentation in industrial facilities. More instrumentation means better visibility into operational and environmental variables that affect overall uptime, safety, and compliance.

WSNs connect critical processes and assets with the systems or experts that can interpret the data or take immediate action. At the end of the day, operational teams with more visibility into their processes can prevent unplanned shutdowns, increase efficiency, and keep workers safe.

WSN is a term used to describe an emerging class of embedded communication products that provide redundant, fault-tolerant wireless connections between sensors, actuators and controllers or systems. WSNs provide access to assets or instruments that were previously deemed unreachable due to physical or economic barriers.

The WSN label typically describes products that provide performance above and beyond traditional point-to-point solutions, particularly in areas of fault tolerance, power consumption and installation cost.

Wireless Challenges While wireless provides clear cost and flexibility advantages, it also presents some challenges. Point-to-point radio communication links are notoriously variable and unpredictable.

A link that is strong today may be weak tomorrow due to environmental conditions, new obstacles, unanticipated interferers and myriad other factors. These factors can be boiled down into three major failure modes: interference, changes in the physical environment that block communication links, and loss of individual nodes.

RF interference: The small portion of the electromagnetic spectrum devoted to general-purpose wireless communication devices is crowded with traffic from Wi-Fi networks, cordless telephones, bar-code scanners, and innumerable other devices that can interfere with communications. Because there is no way to predict what interferers will be present in a facility at a given location, frequency, and time, a reliable network must be able to continually sidestep these interferers on an ongoing basis.

Blocked Paths: When a network is first deployed, wireless paths are established between devices based on the immediate RF environment and available neighbors. Unlike wired networks, these variables often change; paths may later be blocked by new equipment, repositioned partitions, delivery trucks, or very small changes in device position.

Assuring reliability for the life of the network, not just the first few weeks after installation, requires continually working around these blockages in a transparent, automatic fashion.

Node Loss: Node loss is an important issue to consider with wireless sensor networks. While node failure because of semiconductor or hardware malfunction is rare, nodes may be damaged, destroyed or removed during the life of the network.

Additionally, power surges, blackouts, or brownouts can cause nodes to fail unless they have an independent power source. End-to-end reliability requires the networking intelligence that routes around the loss of any single node.

Any of these problems will bring down a point-to-point wireless link. However, with a network architecture designed to protect against these issues, the network can isolate individual points of failure and eliminate or mitigate their impact, allowing the network as a whole to maintain very high end-to end reliability in spite of local failures.

Similarly, a well-designed wireless network architecture will transparently adapt to changing environments, allowing long-term operation with zero-touch maintenance.

WSNs aim to overcome these challenges by applying self-organizing and self-healing intelligence to continuously adapt to unpredictable conditions. The goal of WSN technology is to provide extremely high reliability and predictability for years at a time without constant tuning by wireless experts.

Time Synchronized Mesh Protocol (TSMP) provides a mechanism for WSN intelligence. By defining how a wireless node utilizes radio spectra, joins a network, establishes redundancy and communicates with neighbors, TSMP forms a solid foundation for WSN applications.

TSMP Overview TSMP is a media access and networking protocol that is designed specifically for low power, low-bandwidth reliable networking. Current TSMP implementations operate in the 2.4 GHz ISM band on IEEE 802.15.4 radios and in the 900 MHz ISM band on proprietary radios.

TSMP is a packet-based protocol where each transmission contains a single packet, and acknowledgements (ACKs) are generated when a packet has been received unaltered and complete. Mechanisms are in place to transport packets across a multi-hop network as efficiently and reliably as possible. All measures of reliability and efficiency are done on a per-packet basis.

Packet Structure TSMP packets consist of a header, a payload and a trailer. Packets contain fields that identify the sending node, define the destination, ensure secure message transfer and provide reliability and quality of service information. For the purposes of this article we will discuss the implementation of TSMP on IEEE 802.15.4 radios.

The IEEE 802.15.4 standard specifies a maximum packet size of 127 B, TSMP reserves 47 B for operation, which leaves 80 B for payload.

Time Synchronized Communication All node-to-node communication in a TSMP network is transacted in a specific time window. Commonly referred to as Time Division Multiple Access (TDMA), synchronized communication is a proven technique that provides reliable and efficient transport of wireless data.

Unlike wired systems where nodes can be directly connected by a dedicated wire (media), to the exclusion of neighbors, in a wireless system all devices within range of each other must share the same media. Several other Media Access Control (MAC) mechanisms are available including CSMA, CDMA and TDMA. TSMP is based on TDMA.

Timeslots and Frames In TSMP, each communication window is called a timeslot. A series of timeslots makes up a frame, which repeats for the life of network. Frame length is counted in slots and is a configurable parameter. In this way a particular refresh rate is established for the network.

A shorter frame length increases refresh rate, increasing effective bandwidth and increasing power consumption. Conversely, a longer frame length decreases refresh rate, thereby decreasing bandwidth and decreasing power consumption.

A TSMP node can participate in multiple frames at once allowing it to effectively have multiple refresh rates for different tasks. The concept of slots and frames is illustrated in Figure 1.

Figure 1. TSMP Timeslots and Frames.Synchronization A critical component of any TDMA system is time synchronization. All nodes must share a common sense of time so that they know precisely when to talk, listen, or sleep. This is especially critical in power-constrained applications like WSNs where battery power is often the only option, and changing batteries can be costly and cumbersome.In contrast to beaconing strategies employed by other WSN implementations, TSMP does not begin each frame with a synchronization beacon. Beaconing strategies can require long listen windows which consume power.Instead, TSMP nodes maintain a precise sense of time, and exchange offset information with neighbors to ensure alignment. These offset values ride along in standard ACK messages and cost no extra power or overhead.A common sense of time enables many network virtues: bandwidth can be pre-allocated to ensure extremely reliable transmission and zero self-interference; transmitting nodes can effectively change frequencies on each transmission and the receiving node can keep in lock-step; bandwidth can be added and removed at will in a very predictable and methodical way to accommodate traffic spikes; and many others.

Duty Cycling It is important to note that TSMP nodes are only active in three states: 1) sending a message to a neighbor, 2) listening for a neighbor to talk, and 3) interfacing with an embedded sensor or processor.

For all other times the node is asleep and consuming very low power. In a wireless device the majority (generally >95%) of the total power budget is consumed by the radio, not the processor.

To achieve low power, it is clear that one must minimize radio on time. TDMA is very good at this. Timeslots are measured in milliseconds and in typical WSN applications this leads to a duty cycle of less than 1% for all nodes in the network (including those relaying messages for neighbors).

Because all nodes (including those often called .routers.) can be aggressively duty cycled, TDMA is the only practical solution for a fully battery-powered network.

In addition to slicing the wireless media across time, TSMP also slices it across frequency. This provides robust fault tolerance in the face of common RF interferers as well as providing a tremendous increase in effective bandwidth.

Commonly referred to as Frequency Hopping Spread Spectrum (FHSS), hopping across multiple frequencies is a proven way to sidestep interference and overcome RF challenges with agility rather than brute force.

Another technique to overcome RF challenges is Direct Sequence Spread Spectrum (DSSS). DSSS provides a few dB of coding gain and some improvement in multi-path fading. While beneficial, DSSS is not sufficient in the face of common interferers in the band, including Wi-Fi equipment, two-way radios or even Bluetooth (see figure 2 below).

Figure 2. Frequency Hopping vs. DSSS in 802.15.4 Networks. It should be noted that a combination of FHSS and DSSS provides both interference rejection (FHSS) and the coding gain (DSSS).The other technique for overcoming interference is increasing the radio power, effectively turning up the volume. Although often effective, turning up the volume on IEEE 802.15.4 radios kills battery life and is not an ideal solution for low-power WSNs.Hopping Sequence Upon joining a network, a TSMP node (call it node C) will discover available neighbors and establish communication with at least two nodes already in the network, call them parent A and parent B (more on this in later sections).During this process node C will receive synchronization information and a frequency hopping sequence from both parent A and parent B. The IEEE 802.15.4 standard specifies 16 distinct frequency channels within the 2.4000-2.4835 MHz ISM band . so let’s use 16 as our number. The hopping sequence is a pseudo-random sequence of all available channels. For example the sequence may be: 4,15,9,7,13,2,16,8,1,etc.

Node C receives a distinct start point in the sequence from each parent, and when a new node joins it, it will in turn give a distinct start point to this new child node. In this way each pair-wise connection is ensured to be on a different channel during each timeslot enabling broad use of the available band in any one location.

In operation, each node-to-node transmission (say C to A) is on a different frequency than the previous transmission. And should a transmission be blocked, the next transmission will be to an alternate parent (C to B) on a different frequency. The result is simple but extremely resilient in the face of typical RF interferers.

Posted in Time Division Multiple Access (TDMA), Media Access Control (MAC), IEEE 802.15.4, Time Synchronized Mesh Protocol (TSMP), Bluetooth, Wireless sensor networks (WSNs), Wi-Fi | No Comments »