How to set up an IoT network to get associated data from the sensor to the Cloud? – part III

Thomas Steen Halkier – CEO of NeoCortec, and Zoltan Kiss – Head of R&D at Endrich Bauelemente Vertriebs

In this paper we continue discussing the different possibilities to collect sensor data and get them into the Cloud using the E-IoT ecosystem, which has recently been using NeoCortec’s revolutionary NeoMesh protocol for having an ad-hoc, real low-power, sub-GHz mesh WLAN to collect data locally and gateway it to the Internet from a single access point.

Wireless networks – reliability factors to consider

It is often said that there are no solutions, more reliable in communication, than cables. But cables suffer from potential damages such as breaking, collecting noises. The installation is also usually costly and sometimes not even possible, for instance when the network organization is subject to frequent changes. In systems where high reliability is a must, redundant communication channels are required, which calls for double cable installation, so cabled solutions’ reliability comes with a cost. Offering all advantages of wireless networking, NeoMesh is a cable-like reliable solution which guarantees much higher level of scalability and lower costs.

To understand key reliability aspects of wireless networking, let us look at the performance indicators of such networks. Noise in wireless networks can significantly affect communication performance in both ISM (Industrial, Scientific, and Medical) and licensed bands. In the unlicensed and often crowded ISM bands, interference and non-cooperative transmission can lead to increased noise levels, which degrades the overall signal quality.

On the other hand, even in licensed bands where specific frequencies are allocated to specific users, external factors such as environmental conditions and interference from neighboring bands can cause noise, necessitating effective noise reduction strategies for reliable wireless communication. Static obstacles such as buildings and walls can attenuate and block wireless signals, leading to signal degradation and reduced coverage. Dynamic obstacles such as moving vehicles or people introduce additional challenges with signal interference and fluctuations affecting the reliability of wireless communication in real-time scenarios.

Cyclic redundancy check

In unidirectional communication, where there is no place for handshake and thus no chance for data acknowledgement, the reliability dramatically decreases compared to bidirectional networks with ACK/NAK handshake. To validate the data being received, the most commonly used method is the cyclic redundancy check (CRC). It is an error-detection technique used in digital communication to verify the integrity of data. It involves appending a fixed-size check value to the data, which is calculated based on the remainder of polynomial division. The recipient can use this check value to identify and correct errors that may have occurred during transmission. CRC is particularly effective in detecting burst errors referring to consecutive bits that are corrupted during transmission. CRC, by its nature of polynomial division, is capable of detecting burst errors because it relies on the fact that errors close to each other in the data stream will result in distinct and detectable patterns in the remainder of the division process, allowing for efficient error detection and correction.

CRC16 – effectiveness and limitations

CRC16 is generally effective in detecting errors, including longer random errors, within the limits of its design. The effectiveness of CRC in detecting errors is determined by the size of the CRC, which in the case of CRC16 is 16 bits. This means that it can detect errors of up to 16 bits in length. However, it´s important to understand the limitations of CRC. While it is robust for many applications, it is not foolproof, and there is always a small possibility that certain types of errors may go undetected. Additionally, CRC is more effective at detecting burst errors rather than random errors.

If the requirement is to detect longer random errors with higher certainty, a larger CRC or other error-detection and correction codes with greater capabilities may be considered. The choice of an error-checking method depends on the specific requirements of the application and the desired level of error detection and correction.

We can however say that CRC16 is able to detect longer random errors with a probability of 99,998474%. This means that payload messages with bit errors received may be marked as "ok" by the CRC error check algorithm. In case of thousands of payload packages delivered each day in a large network, at least a few packages will be accepted as "good" although they are "bad". If the CRC checksum is as high as 32 bits, the undetectable burst error length will be longer, and the probability of blocking such packages will be 99,99999997671%, significantly better than CRC16, and with a nearly-zero chance to let errors through.

Possible causes for link breaking

In a point-to-point topology, or even in a star network topology, if the link between the sending device and the receiving device is no longer reliable, the communication breaks down – there is no alternative path for the data to flow. The link between the two devices can be broken because of noise, or because it is obstructed – either temporarily or permanently, by any obstacles. The noise level will vary from location to location, and also over time. This means that even though a system installed in one location may work flawlessly, it may not work as reliably in another location. Similarly, a system may work with no issues at the time of its installation, but sometime later, problems may start to occur. That could be due to other installed systems which operate in the same frequency band.

To be continued in the next issue.

www.endrich.com