.png){kind=icon}
.png){kind=icon}
Return
DRL-Based Optimization Algorithm for Wireless Powered IoT Network
.png){kind=icon}
PRE
AI
DOI：10.1007/978-981-97-0808-6_11.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
The extensive applications of Internet of Things (IoT) bring development and prosperity to multiple industries and greatly influent all aspects of human lives, which rely on the timely computation and communication. However, the limited battery and computing capacity of the common IoT units cannot satisfy the need of processing computation intensive and time sensitive tasks. As a result, the combination of wireless power transmission (WPT) and mobile edge computing (MEC) provides a promising approach of dealing with aforementioned problems. The Hybrid Access Point (HAP) transmits radio frequency (RF) energy to provide power transmission for wireless devices (WDs), and processes tasks offloaded from WDs with the equipped edge computing servers (ECSs). In this paper, we consider a wireless powered mobile edge computing network containing an HAP, multiple WDs. By jointly optimizing energy transmission time duration, partial offloading decisions and transmission power allocation strategy, we obtain the maximum of the sum computation rate (SCR). Firstly, we formulate this as a non-convex problem which is hard to address. Secondly, we decompose the problem into a top-problem of optimizing the energy transmission time duration and a sub-problem of optimizing the partial offloading decisions and transmission power allocation strategy. Finally, we design a DRL-based offloading algorithm, which applies a DNN network with exploring and updating strategy, to address the top-problem and propose an effective optimizing algorithm to address the sub-problem. Extensive numerical results reveal that the proposed algorithm reaches near-optimal SCR and greatly reduces time latency and complexity compared to benchmark algorithms.
Keywords:
Mobile-edge computing
wireless power transfer
reinforcement learning
partial offloading
resource allocation
Journal
A
IF:
0
Papers:
11
Citations:
0