SF-Partition-Based Clustering and Relaying Scheme for Resolving Near–Far Unfairness in IoT Multihop LoRa Networks

Abstract

Long range (LoRa) is one of the most successful low-power wide-area networking technologies because it is ideally suited for long-distance, low-bit rate, and low-power communications in the unlicensed sub-GHz spectrum utilized for Internet of things (IoT) networks. The effectiveness of LoRa depends on the link budget (i.e., spreading factor (SF), bandwidth (BW), and transmission power (TX)). Due to the near–far effect, the allocation of a link budget to LoRa devices (LDs) in large coverage regions is unfair between them depending on their distance to the GW. Thus, more transmission opportunities are given to some LDs to the detriment of other LD’s opportunities. Numerous studies have been conducted to address the prevalent near–far fairness problem. Due to the absence of a tractable analytical model for fairness in the LoRa network, however, it is still difficult to solve this problem completely. Thus, we propose an SF-partition-based clustering and relaying (SFPCR) scheme to achieve enormous LD connectivity with fairness in IoT multihop LoRa networks. For the SF partition, the SFPCR scheme determines the suitable partitioning threshold point for bridging packet delivery success probability gaps between SF regions, namely, the lower SF zone (LSFZ) and the higher SF zone (HSFZ). To avoid long-distance transmissions to the GW, the HSFZ constructs a density-based subspace clustering that generates clusters of arbitrary shape for adjacent LDs and selects cluster headers by using a binary score representation. To support reliable data transmissions to the GW by multihop communications, the LSFZ offers a relay LD selection that ideally chooses the best relay LD to extend uplink transmissions from LDs in the HSFZ. Through simulations, we show that the proposed SFPCR scheme exhibits the highest success probability of 65.7%, followed by the FSRC scheme at 44.6%, the mesh scheme at 34.2%, and lastly the cluster-based scheme at 29.4%, and it conserves the energy of LDs compared with the existing schemes.

A Decentralized Dynamic Relaying-Based Framework for Enhancing LoRa Networks Performance

“…In [21], a partitioning technique based on SF is proposed to achieve fairness in multihop LoRa networks. Nodes are clustered into high and low SF groups, with the GWs and relays selecting clusters using the naive Bayes algorithm.…”

“…Intermediate GWs are distributed in rings based on the Fibonacci distance propagation model, allowing efficient packet forwarding to the primary GW. In [26], an implicit overhearing node-based multi-hop communication (IOMC) scheme utilizes relay nodes performing overhearing and forwarding to adhere to duty cycle limits. A synchronization scheme, was proposed in [27] to coordinate transmissions by a central entity by encoding timeslot assignments using a probabilistic data structure.…”

Implicit Overhearing Node-Based Multi-Hop Communication Scheme in IoT LoRa Networks

“…The authors propose a TDMA protocol that assumes the communication is only between the nodes and a sink and uses beacons to synchronize the nodes in each epoch, thus benefiting from the concurrent transmission feature of LoRa to decode correctly at most one packet in each slot and receiver. Works in [12,27,28] presented a forwarding relay node and clustering approach to not only enhance coverage and concurrent transmission of LoRa networks using multi-hop communication but also make the system more energy efficient. In addition, devices are categorized into several clusters with the motive of streamlining the operations of the network to maximize energy efficiency and consequently prolong the network lifetime.…”

“…To do this, we exploit a ranking system to evaluate sensor nodes in the OH candidate set. The ranking system calculates the rank of a candidate OH node OH i in the OH candidate set as Equation (27).…”

“…For applying the normalized values for both BER and the residual energy, we use L BER(max) and L BER(min) for BER and E OH i ,re(max) and E OH i ,re(min) for the residual energy in Equation (27). By the ranking system, when a candidate OH node OH i has the highest rank as (R(OH i )) in Equation ( 28), it becomes the best OH node.…”