Reinforcement Learning-Based Near-Optimal Load Balancing for Heterogeneous LiFi WiFi Network

Abstract

Owing to the non-overlapping spectrum, Light fidelity (LiFi) and WiFi technologies can coexist and form a heterogeneous LiFi WiFi network (HLWN). The performance of HLWN significantly depends upon the load balancing strategies. Since load balancing of HLWN is a non-convex mixed-integer nonlinear programming (MINLP) optimization problem, it is mathematically intractable, and therefore, the conventional optimization methods fail to provide an optimal global solution. Although an optimal solution can be obtained using the exhaustive search method, it would be computationally complex. Therefore, in this paper, a reinforcement learning (RL) based algorithm is explored for solving the load balancing problem for the downlink HLWN at reasonably low complexity and near optimal performance. We have proposed three different reward functions for RL; the first and second reward functions work toward maximizing average network throughput and user satisfaction, respectively. The third reward function is designed to maximize the long-term system throughput and ensure at least 50% user's satisfaction for all users. In order to study the effects of link aggregation on system performance, this work considers two different types of receiver schemes, namely, single access point (SAP) and link aggregation (LA) scheme. While the SAP allows the user to receive data only from a single AP, the LA scheme allows the user to receive data simultaneously from both LiFi and WiFi AP. This paper also includes effect of random orientation of the receiver device and handover overhead. Further, concepts of domain knowledge have been included in this work to reduce the computational complexity of the algorithm. The proposed system performance is compared with the two benchmarks: received signal strength (RSS) and exhaustive search based on the computational complexity, average system throughput, and user satisfaction. It is shown that the proposed RL scheme outperforms the RSS scheme in average system throughput and user satisfaction. The RL scheme with an appropriate reward function provides a matching performance to the exhaustive search at reasonably low complexity.

Integration of Network Slicing and Machine Learning into Edge Networks for Low-Latency Services in 5G and beyond Systems

“…Secondly, optimizing the synchronization process for heterogeneous networks is indeed a long-term decision that is affected by multiple conditions. Such problems are considered decision-making tasks where RL offers fast near-optimal solutions; this is also validated by other research in the literature [63,64].…”

Exploration of Agricultural Economic Management Methods under Internet + Mode

“…ZigBee network protocol includes physical layer (PHY), media access control layer (MAC), network layer (NWK), and application support layer (APS) structure. e bottom layer is composed of the physical layer and the media access control layer, which meets IEEE802.15.4 standard protocol and can be directly connected with the RF transceiver [14]. e network layer is concerned with the establishment and maintenance of the entire network.…”

Sequential Load Balancing for Link Aggregation Enabled Heterogeneous LiFi WiFi Network

“…Limited work has been done on LA enabled HLWNs [13]- [20], which has been summarized in Table 2. In [13]- [16], authors have implemented LA in HLWN at different layers of the protocol stacks and demonstrated proof-of-concept.…”

“…However, as they were dealing with real-time industrial control applications, their focus was to guarantee transmission reliability and latency requirements. In our previous work [20], we used a centralized reinforcement learning algorithm to determine the user association in LA enabled HLWN and hybrid LiFi WiFi network. In that article, we focused on determining the best reward function for user association to ensure a particular user satisfaction and high data rate.…”

“…r A more realistic framework with OFDMA multi-user access, adaptive modulation and coding, heterogeneous users, handover overhead and LA overhead has been r The proposed method is compared with the RSS and exhaustive search method with equal resource allocation and RSS with optimum resource allocation, as explained in section IV-A and IV-B. Furthermore, the proposed method has also been compared against the algorithm proposed in [20]. The performance of all these schemes is evaluated based on their computational complexity, average network throughput, and user satisfaction.…”