Abstract Internet Protocol version 6 (IPv6) over low power wireless personal area networks (6LoWPANs) forms a majority of traffic share in Internet of things (IoT) where quality of service (QoS) becomes obligatory for multitude of sensor inputs. 6LoWPANs are interference prone due to the fact that the data link and physical layers utilize the IEEE 802.15.4 standard for communication. Interference in 6LoWPANs results in poor QoS in terms of packet reception ratios and packet loss rates and also in poor network stability and reduced network lifetime. A deep neural network based routing algorithm is proposed which offers multiple solutions to the interference problem and selects the best solution in order to reduce interference. The proposed routing algorithm improves the network lifetime, delay and jitter on an average by 50%, 40%, and 25% respectively compared to the standard 6LoWPAN routing protocol. The signal to interference and noise ratio is also improved on an average by 18 decibel.

中文翻译：

---

使用深度神经网络优化 6LoWPAN 中的网络寿命和 QoS

摘要 低功率无线个域网 (6LoWPAN) 上的互联网协议版本 6 (IPv6) 形成了物联网 (IoT) 中的大部分流量份额，其中服务质量 (QoS) 成为众多传感器输入的必要条件。由于数据链路和物理层使用 IEEE 802.15.4 标准进行通信，因此 6LoWPAN 容易受到干扰。6LoWPAN 中的干扰会导致在数据包接收率和数据包丢失率方面的 QoS 差，以及网络稳定性差和网络寿命缩短。提出了一种基于深度神经网络的路由算法，该算法为干扰问题提供多种解决方案，并选择最佳解决方案以减少干扰。所提出的路由算法将网络寿命、延迟和抖动平均提高了 50%、40%、与标准 6LoWPAN 路由协议相比，分别为 25% 和 25%。信干噪比也平均提高了 18 分贝。