Looking before Crossing: An Optimal Algorithm to Minimize UAV Energy by Speed Scheduling with a Practical Flight Energy Model

Shan, Feng; Luo, Junzhou; Xiong, Runqun; Wu, Wenjia; Li, Jiashuo

doi:10.1109/infocom41043.2020.9155376

Cited by 33 publications

(19 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this paper, unlike [13]- [15], we consider a more realistic scenario where the UAV is equipped with a battery that has limited energy storage. It is worth mentioning that as the energy consumption for wireless data collection process (i.e., broadcasting the wakeup signal and receiving data packets) is much lower than that of the flying operation [7], we only focus on optimizing energy consumption for the UAV's flying operation in this work. In a time slot, we assume that the UAV's velocity is constant (as in [13], [17]), but in different time slots the UAV can choose to fly at different speeds, e.g., 𝑣 𝑤 = {𝑣 1 , .…”

Section: System Modelmentioning

confidence: 99%

“…Each speed may cost a different amount of energy. For example, if the UAV flies faster, it may use more energy per time slot [7].…”

Section: System Modelmentioning

confidence: 99%

“…The energy usage efficiency can be represented by the cost of choosing a flying speed, i.e., energy consumed by the UAV to fly at speed 𝑎 during a time slot 𝑡. Clearly, the selected speed determines the energy consumption per time slot of the UAV, e.g., a low speed will cost the UAV less energy per time slot than that of a high speed [7]. At time slot 𝑡, the cost of performing action 𝑎 is denoted by 𝑚 𝑎 𝑡 , and the number of collected data packets at the current state 𝑠 is denoted by 𝑑 𝑠 𝑡 .…”

Section: Reward Functionmentioning

confidence: 99%

“…Conversely, at other zones that contain less or even no IoT nodes, the UAV can fly faster to avoid missing opportunities from other active zones. Note that given a fixed flight time, a high speed may cost the UAV more energy than a low speed [7]. Therefore, controlling the UAV's speed can also affect the UAV's energy consumption.…”

Section: Introductionmentioning

confidence: 99%

“…However, in practice, a UAV can choose different speeds during its data collection process depending on its surrounding environment. Alternatively, the UAV's speed can strongly influence the system's efficiency because it has a substantial impact on energy consumption during the data collection process [7]. Thus, optimally controlling UAVs' speed can significantly improve the energy usage and data collection efficiency of the system, especially in UAV-assisted IoT data collection networks where UAVs have limited battery capacities.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Joint Speed Control and Energy Replenishment Optimization for UAV-assisted IoT Data Collection with Deep Reinforcement Transfer Learning

Chu¹,

Hoang²,

Nguyen³

et al. 2021

Preprint

View full text Add to dashboard Cite

devices are distributed at hard-to-reach locations such as on top of trees, outside high-rise buildings, or even under bridges. In this case, cellular base stations can act as data collectors in IoT data collection networks. However, the limited energy storage capacity and communication capability hinder IoT devices from transmitting their data over long distances. Thus, IoT data collection networks have been demanding effective and flexible solutions for data acquisition.Recently, UAVs have been emerging as a promising approach to tackle the above challenges. In particular, when UAVs act as on-demand flying APs, thanks to their aerial superiority, they can establish good line-of-sight (LoS) links for the IoT nodes. This leads to better wireless communication channels, and thereby improving the quality of service (QoS) in comparison with traditional approaches, especially for IoT applications that are sensitive to delay and/or require stable communications [3]. In addition, in remote areas without access to terrestrial infrastructures, UAVs can provide a much more economic solution to collect IoT data than traditional approaches, e.g., long-range ground broadcasting stations or high-cost satellite communications. Another important advantage of UAVs is that they can be promptly established in emergency circumstances, in which the existing infrastructure is disrupted and incapable of receiving data from IoT devices [4]. Due to the flexibility, mobility, and low operational cost, UAVs have been being deployed as flying APs for some real-world projects, e.g., Google's Loon and Facebook's Aquila [5], [6].However, there are still some challenges that hinder the applications of UAVs in IoT data collection networks. In particular, unlike traditional solutions for collecting IoT data (e.g., deploying fixed APs), UAVs have limited energy resources supplied by batteries. When the UAVs' batteries are depleted, they must replenish their energy by flying back to the charging stations to charge or replace their batteries. It is worth noting that given a fixed working duration, the more time the energy replenishment process takes, the less time the UAVs can spend for collecting IoT data. Alternatively, the energy replenishment process is highly dynamic since it depends on the distance between the UAV and the charging stations. Therefore, optimizing energy usage and the energy replenishment process is critical to achieving high system performance, but very challenging in practice. Moreover, the UAVs often fly around to collect IoT data, while IoT nodes are statically allocated over different zones, and their sensing data are random depending on surrounding environments. To that end, optimizing operations of UAVs in different zones to

show abstract