Handling Coexistence of LPWAN with Other Networks
Abstract
With its capability to enable low-power (milliwatts) wireless communication at low data rates (kbps) over long distances (kilometers,), the Low-Power Wide-Area Network (LPWAN) technology represents the next frontier of communication in various military applications, (e.g., enemy tracking, battlefield monitoring, and surveillance systems) that can extend over large areas. However, the rapid grow,th of LPWANs in the limited spectrum brings forth the challenge of coexistence of many networks and devices in the same band. The im,mediate effect of such coexistence is degraded network performance in terms of throughput, latency, and energy. Some networks or dev,ices may even suffer from spectrum starvation. Repeated attempts to access the spectrum will drain their batteries. In Naval applica,ped to handle the impending challenge of coexistence. Their nodes have verylow computation power, memory, and energy typically suppl,range wireless sensor network will not work for LPWANs. In massive crowds of coexisting networks, the interference pattern can be ha,rd to detect for LPWAN nodes. Due to long range, they are subject to an unprecedented number of hidden nodes, requiring new techniq,ues for combating such interference. In this project, we specifically address the coexistence problem for SNOW (Sensor Network Over,White spaces) so that it can operate in the presence of many other coexisting networks/devices. To avoid the cost of the licensed ba,nd and the crowd of the limited ISM band, we designed SNOW as an LPWAN architecture to support scalable wide-area Internet-of-Things, applications over the widely available TV white spaces (unused TV band). Our preliminary work showed advantages of SNOW over other, LPWANs in scalability and energy. It allows a base station (BS) to receive concurrent transmissions made by the nodes asynchronousl,y. It also allows concurrent downlink communication. This is done through splitting a wide spectrum into many narrowband orthogonal,subcarriers enabling parallel data streams to/from numerous distributed nodes from/to the BS. Following are the key contributions,we shall make through this project. (1) First, we propose the design of a novel embedded learning agent at SNOW nodes based on a li,ghtweight Reinforcement Learning (RL) to improve the performance of a SNOW under coexistence with many independent networks. An RL a,gent (i.e., a node) is effective in an unknown environment where it learns through its experience. We shall specifically adopt Q-lea,rning which is well-suited for coexistence handling at the SNOW nodes as it entails relatively low computation requirement compared,to other machine learning approaches. (2) To handle a more severe scenario of interference such as jamming, we propose a game-theore,tic approach for frequency hopping at the BS. The uniqueness of this approach lies in designing the players and their actions with r,es,s. We propose to play a static game but repeatedly with belief update and/or reinforcement learning to find the Nash equilibrium(s),, thus identifying the effective strategies against jamming. (3) We also propose to integrate the orthogonal variable spreading facto,r technology in SNOW to facilitate the coexistence of multiple SNOWs. Furthermore, we propose an adaptive subcarrier bandwidth appro,ach that enables the nodes to dynamically adjust their bandwidths and select some interference-free part of their subcarriers for co,mmunication under persistent interference. We will implement the proposed ideas on SNOW hardware and evaluate through experiment und,er various coexistence scenarios.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Feb 08, 2022
- Source ID
- N000142212155
Entities
People
- Abusayeed Saifullah
Organizations
- Iowa State University
- Office of Naval Research
- United States Navy