The world is transitioning to the 5G era. 5G involves more than simply upgrading to “better” radios in the network. The Internet of Things (IoT) is part of the biggest plans 5G has for us. The IoT is to have various devices connected such as wearables, appliances, and machines. The research topics of wireless sensor networks (WSN) bring a lot of needed insight to this area. Localization is one such topic.

The paper points out that GPS-based localization is expensive, both in terms of hardware cost and in energy consumption, and the signals are not always available in the deployed environment. As an alternative, node location can be estimated from known nodes or anchor nodes in the network. There are also two general approaches in this latter alternative: range based and range free. Range-based approaches make use of relative location information from the anchors, such as angle/time of arrival and received signal strength indicator. Range-free approaches do not.

The authors propose soft computing techniques in range-free location approximation. These techniques such neural networks, fuzzy logic, support vector machines, and evolutionary computation are more common in machine learning or artificial intelligence. For readers interested in WSN but not very familiar with these techniques, the information in the paper may seem a bit overwhelming. The authors have done a good job formulating the problems and algorithms for readers to follow. After that, the authors show us the estimation error and computation complexity for these techniques through simulation. It would definitely be interesting to see the future research mentioned at the end of the paper. Experiment designs based on TinyOS networks provide a more realistic application of the soft computing theories.