In recent years, the use of the Internet of Things has become very common and has become an inseparable part of human life in many countries. For this reason, this field has attracted the attention of many researchers and large companies. Generally, low-power wireless networks are used in the Internet of Things, where the energy source of its members is a battery. Among the most important criteria for eva‎luating these wireless networks, energy consumption, latency, and throughput can be mentioned, all of which are affected by the packet delivery ratio (PDR) of the network. However, in the current world, achieving a high PDR in the very crowded 2.4 GHz frequency is not an easy task at all. Existing WiFi and Bluetooth networks have created a crowded environment in the 2.4 GHz frequency, making it very difficult to achieve a high PDR, especially for weak networks like Bluetooth. In one study, a method called Blacklist or PDR-Exclusion has been introduced to improve the Bluetooth PDR, which can maintain the PDR at a very good level. However, in another study, it was stated that the blacklist alone is not sufficient in crowded networks, and a method for channel return is necessary. Therefore, this research presents a method called informed exploration for adaptive frequency hopping eAFH. In the present research, additional investigations have been conducted on existing methods, and these investigations have shown how ineffective basic communication methods without a blacklist can be. According to these studies, the average network PDR decreases to about 73% with the presence of only one WiFi channel and even drops momentarily to below 60%. According to the investigations conducted on PDR-Exclusion, this method performs exceptionally well against one WiFi channel, raising the average PDR to about 99.68%. Additionally, this method does not allow the PDR to drop below 95% even momentarily after the transient state. However, further investigations revealed that this method loses its excellent performance with the addition of just one Bluetooth interference. Due to the lack of discovery, in PDR-Exclusion, over time, channels are removed until the number of channels in the the hopping list reduces below a minimum threshold, and then the hopping list is reset. With the resetting of the hopping list, the transient state reoccurs, and the momentary network PDR plummets to about 75%. This also causes the average PDR to decrease to 97.91%. However, eAFH performs well in this environment. This method maintains a minimum momentary PDR of around 90% while also keeping the average PDR at 98.36%. However, this method is not flawless and even in the absence of Bluetooth interference, it performs so many unnecessary discoveries that the average PDR reaches 99.06%. This thesis suggests that discovery is appropriate to prevent the resetting of the hopping list. However, excessive discovery in a high number of channels can lead to a decline in the packet delivery rate. For this reason, it is proposed that the discovery rate be a function of the number of channels so that when necessary, the discovery rate increases compared to normal conditions, and when the number of active channels is high, the discovery rate decreases compared to normal conditions. Various functions were examined for this purpose, and it was found that some of them, including linear functions, do not yield acceptable results. However, according to the investigations, the exponential function delivers good results and can increase the average network PDR and the number of active channels by 0.16% and 1.4 channels, respectively, in the absence of Bluetooth, while also improving their standard deviation. Additionally, in the presence of Bluetooth, this function increases these two metrics by 0.28% and 5, respectively, compared to eAFH, and also improves their standard deviation.

مهدي مهدوي

امير خورسندي كوهانستاني , عارف كريمي افشار