IEEE 802.11ah introduces the restricted access window (RAW) mechanism at the MAC layer to mitigate the effect of collisions and to improve the network performance. In an Internet of Things network with heterogeneous traffic requirements, the challenging aspect of the RAW mechanism is to choose the optimal number of RAW slots (ONRs) and to dynamically assign the RAW slots according to the transmission requirements of the devices. In this article, we propose an optimization model by using adaptive neuro-fuzzy inference system (ANFIS) to find the ONRs. The ANFIS is trained with network size, modulation and coding schemes, and the ONRs found analytically. Further, we propose a dynamic allocation of RAW slots (DARS) algorithm to classify the devices based on their traffic criteria. The proposed algorithm classifies the network using self-organizing maps and dynamically allocates the RAW slots to each group. We present a mathematical model to assess the throughput and energy consumption. The results show that the throughput performance is significantly improved and the energy consumption is considerably decreased by using the ONRs. Further, the DARS scheme effectively provides differentiated services to the group of devices in contrast to the default uniform grouping scheme. It is observed that the proposed optimization scheme significantly improves the throughput by 20% and reduces the energy consumption by 26% for a network of 4000 devices, compared to the legacy RAW mechanism. Further, the DARS scheme increases the average data transferred by the devices with the highest traffic requirements by 68%. Finally, all the analytical findings are validated by simulation studies using ns-3.

IEEE 802.11ah在MAC层引入了受限访问窗口（RAW）机制，以减轻冲突的影响并提高网络性能。在具有不同流量需求的物联网网络中，RAW机制的挑战性方面是选择最佳数量的RAW插槽（ONR），并根据设备的传输要求动态分配RAW插槽。在本文中，我们提出了一种使用自适应神经模糊推理系统（ANFIS）来查找ONR的优化模型。对ANFIS进行了网络大小，调制和编码方案的培训，并通过分析发现了ONR。此外，我们提出了RAW时隙（DARS）算法的动态分配，以根据设备的流量标准对其进行分类。所提出的算法使用自组织映射对网络进行分类，并为每个组动态分配RAW插槽。我们提出了一个数学模型来评估吞吐量和能耗。结果表明，使用ONR可以显着提高吞吐性能，并显着降低能耗。另外，与默认的统一分组方案相比，DARS方案有效地向设备组提供区分服务。可以看出，与传统的RAW机制相比，针对4000个设备的网络，所提出的优化方案可将吞吐量显着提高20％，并将能耗降低26％。此外，DARS方案将流量需求最高的设备传输的平均数据增加了68％。最后，