Reliability of Self-Organizing Wireless Networks

wangshui

Reliability of Self-Organizing Wireless Networks [Copy link]

Many process personnel have heard the term “wireless instrumentation” and thought that’s great, no wires, but how reliable is it? How do you know if you are receiving the right data on time? Is it always the right data? Let’s take a closer look at this question. What are the limits of reliability for self-organizing wireless networks? Where should we apply this technology in a dynamic industrial environment? Are wireless field devices really reliable enough to handle today’s complex industrial environment?
　　The Next Generation Self-Organizing Networks and HART Communication Foundation Wireless Working Groups are working to address these issues. The HART Working Group is focused on protecting the investment of current users. They are adopting a series of powerful technical standards to bridge the more than 15 million smart devices currently installed into the wireless era. At the end of this phase, these technical specifications will be combined with the SP-100 Best Practices Guidelines to further help users take advantage of the digital factory architecture to close the gap with advanced asset management models.
　　This article will focus on the basic principles of reliable data transmission from an industrial wireless field device to the end user. Please note that we will focus on the reliability of wireless communication here. While overall network reliability depends on the sensors, wireless communications, host integration, and data management, a complete wireless network
is only as reliable as its weakest link.
　　At the heart of all wireless network protocols is energy conservation and management. Wireless instruments need to power a radio, sensors, and embedded electronics. Primary batteries that replace wired power can last for years, and for such batteries the wireless network needs to operate in a duty cycle mode where all electronics are turned off (also called "sleep mode") when the sensors are not operating or not communicating.
　　Duty cycling a sensor can be accomplished with simple timing circuits, but how can a network operate in duty cycling mode? Not only do all devices need to wake up at the same time, but they also need to organize and communicate information in a reliable manner.
　　For the sake of discussion, we need to define two terms: network reliability and path reliability. Network reliability is the ratio of information received to information sent between the gateway and wireless devices in the network. This is a measure of the ability of the end user to receive the data they need from the field device. Associated with this term is “network latency,” which is the amount of time it takes for information to be received by a host computer after it has been wirelessly transmitted from a field device.
　　Path reliability refers to the ratio of messages received to messages sent between two wireless devices in the network, or between a wireless device and a gateway.
　　Star topologies (with the limitations of traditional point-to-point/line-of-sight solutions) typically achieve only 40% path reliability, which is clearly far from ideal. For ad hoc networks, typical path reliability is almost the same as traditional line-of-sight technology. But there is a key difference. By having wireless devices act as intermediate routers, ad hoc networks are able to achieve network reliability exceeding 99%. When information cannot be transmitted through a particular channel, ad hoc networks can provide alternative channels (redundancy). The key to achieving such excellent reliability for ad hoc networks is the use of multiple neighboring devices within hundreds of feet to achieve message transmission redundancy.
　　The example in Figure 1 shows that the reliability of channels AB, AC, and AD are all less than 99%, but the probability is that one channel will be open at any time when it is needed.


Figure 1: Example of a self-organizing network showing how channels with less than 99% reliability can be combined into reliable communication paths.

The vendor of the self-organizing network equipment will differentiate itself based on the automation capabilities of the equipment in part to efficiently decide which paths are open. A well-managed network should be able to identify that paths AB (65%) and AD (85%) are more reliable, while path AC (40%) should be used as a backup path. This capability allows the network to quickly self-adjust to the changing industrial environment, forming a scalable network structure that can easily cover the entire process unit. On the other hand, this eliminates the need for expensive and unreliable site surveys. And because the industrial environment is constantly changing, the results of such site surveys only reflect the situation at a certain moment.
　　For example, if path AB is blocked by a turning truck, scaffolding or other interference sources, the wireless device will first use paths AC and AD. The network will identify the current best path and automatically configure itself to optimize the communication path. This happens both when the network is installed and when the network operates in a changing environment over a long period of time. This capability allows owners to relax because it is the only network technology that can achieve the ideal "just work out of the box". Experience has shown that if this self-organizing network is used to connect standard industrial sensors and information systems, even the most inexperienced factory technicians can install and operate it.
　　When you need to communicate, you must synchronize the wireless devices. There are two different principles to do this: TDMA (time division multiple access) and CSMA (collision sense multiple access). For TDMA, each wireless device knows exactly when it will communicate within the network and also knows the number of communications per unit time. From the example in Figure 1, we can see that device A will transmit data to device B at the scheduled time.

Device B will then store the data and transmit it to another device or directly to the gateway at a scheduled time. The main advantage of this approach is that each message has a specific time period to be transmitted through the network, thus reducing the problems caused by message collisions. It allows various wireless technologies and solutions to coexist in the same factory, which gives us the most reliable way to achieve truly scalable wireless networks in a factory with thousands of devices. And all devices in this type of wireless network do not need to be active at the same time, which can greatly extend the battery life of the devices. For the CSMA method
　　, all devices in the network must be active and try to communicate at the same time, not just a pair of devices. From the example in Figure 1, we can see that if the information from device B and device C arrives at the gateway at the same time, they will "collide" and then they will try to transmit the information again through different channels. Therefore, the main disadvantage of CSMA networks is that it takes a lot of power to achieve scalability because resending the "colliding data" requires additional communication. This situation has a great negative impact on battery life. When the number of wireless devices in the network increases, the number of collisions will also increase accordingly. For a given duty cycle window, when all devices are communicating, only a limited amount of intelligence can
get through before the communication channel is blocked by repeated retransmission activity. For this reason, network reliability and response time are unpredictable, and only minimal networks can be supported by primary batteries. If the CSMA network is actively powered through the wire, it can retransmit infinitely (with a long response time), but this loses the advantages of a true wireless solution.
　　Another aspect of reliability is to ensure that the communicating devices have a clear frequency channel to successfully complete the information transmission. With the increasing use of wireless technologies in factories (such as RFID, Wi-Fi systems, etc.), it is very likely that the communication frequencies of wireless devices will be blocked. This requires the use of frequency agility.
　　The most common form of frequency agility is frequency hopping, in which a pair of adjacent wireless devices will repeatedly try new frequencies until they find a clear channel for communication. This technology greatly increases the ability of many wireless devices using different technologies (RFID, Wi-Fi and other wireless technologies) to coexist.
　　For TMDI networks, the process of determining which frequency to communicate on is already included in the synchronization and advanced network management algorithms. Given that the time required to transmit information between adjacent devices is measured in milliseconds, and that networks can operate at many different frequencies, multiple ad hoc networks can easily coexist in the same process unit.
　　With CSMA networks, all wireless devices have to communicate simultaneously on the available frequencies, which reduces network reliability due to message collisions, long response times, and high device power consumption.
　　The main obstacle to wireless networks is power consumption, which limits the life of device batteries. Unless users are willing to replace batteries every few months or introduce hard wiring to power wireless devices, the update rate of wireless device measurement data will be significantly lower than that of wired systems to maximize the time between maintenance. The
　　first generation of ad hoc network solutions will target applications that require data updates approximately every sixty seconds. This is a completely different concept from wired systems, where data refresh rates in the millisecond range are common.
　　Although not currently feasible for high-speed closed-loop control, a secure ad hoc network with a 60-second data refresh rate can accommodate typical monitoring applications, open-loop control applications, and certain closed-loop control loops that do not require high response times.
　　Any plant with hundreds, if not thousands, of monitoring points that are monitored by operators, either manually or by rounds. These applications are ideal opportunities for self-organizing networks. Isn't a minute-long data update fast enough for monitoring applications? A 60-second data refresh rate is nearly real-time, compared to once per shift, once per day, once per week, once per month, or never. Further, this eliminates manual note-taking errors, dial reading errors, and poor repeatability when using handheld measuring devices.
　　Many potential open-loop applications include those that may take an operator a long time to obtain the proper work permit or require an operator to go out to perform the appropriate "control action," such as going out to start a pump or opening/closing a manual shutoff valve. In these cases, a one-minute refresh time is a fraction of the time it takes an operator to go out to the appropriate location.
　　The key to controlling the response time of a self-organizing network is to ensure that the selected vendor network has a time stamp on each data acquisition from each wireless device and can provide confirmation of successful transmission. Reliable time tags can compensate for the network's reaction time, and confirmation of successful transmission will verify that the information has indeed reached the host.
　　As energy efficiency continues to improve and management is strengthened, the ability of self-organizing networks to increase measurement and transmission speeds will make more non-safety-related/non-critical applications more flexible.
　　By then, there will really be hundreds or even thousands of measurement points that can be easily integrated into a typical factory system. This will greatly improve users' ability to optimize assets and make them more competitive in the global market.