IoT Spatial Resource Identification Using Reinforcement Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing IoT networks face inefficiencies in spectrum utilization due to unused spatial frequency resources, particularly in femtocell networks, leading to suboptimal performance for indoor and cell boundary IoT devices.
Innovation Solution
A system and method utilizing machine learning, specifically reinforcement learning, to identify and optimize spatial resources for device-to-device communication, enabling efficient spectrum utilization by IoTDs through a radio resource harvesting edge (RRHE) that collects and optimizes identification parameters for IoTDs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If macro base stations are used for wireless access, then network coverage is provided, but signal quality for indoor and cell boundary IoT devices deteriorates
Solution Approach 1:
The network is segmented into macro base stations for coverage and femto base stations for localized enhancement. Femto base stations are deployed in specific locations (indoor environments and cell boundaries) to provide localized high-quality access, dividing the network function into hierarchical layers that address different spatial requirements
Solution Approach 2:
The solution transitions from a two-dimensional macro base station coverage model to a three-dimensional hierarchical architecture by introducing femto base stations at lower levels (indoor deployments) and specific spatial positions (cell boundaries), adding vertical and localized spatial dimensions to the network coverage
2Productivity
If femto base stations are deployed to improve signal quality, then spectrum sharing efficiency improves, but unused spatial frequency resources remain
Solution Approach 1:
The system implements feedback mechanisms where femto base stations and IoT devices report channel state information, interference levels, and resource utilization metrics. This feedback enables the network to dynamically identify unused spatial frequency resources and adjust resource allocation to eliminate waste while maintaining efficient spectrum sharing
Solution Approach 2:
The system dynamically changes spatial resource allocation parameters based on real-time network conditions. By adjusting beamforming parameters, spatial multiplexing configurations, and frequency resource assignments, the system identifies and utilizes previously unused spatial frequency resources while maintaining efficient spectrum sharing
3Productivity
If machine learning is applied to identify spatial resources, then spectrum efficiency improves, but system complexity increases
Solution Approach 1:
The patent introduces a spectrum resource identification module as an intermediary that handles the complex machine learning operations. This module acts as a mediator between the physical layer resources and the network layer, encapsulating the complexity of spatial resource identification algorithms while providing simplified interfaces to other network components
Solution Approach 2:
The system implements self-service mechanisms where the machine learning model automatically identifies and allocates spatial resources without manual intervention. The model continuously learns from network conditions and autonomously optimizes resource allocation, reducing the need for complex manual configuration and management
Data Source
AI summary
Disclosed is a system for identifying a spatial resource for an IoT network, the system including a plurality of IoTDs (IoT devices) configured to identify an empty spatial resource using an identification parameter and provide spectrum identification information corresponding to a result thereof, and a radio resource harvesting edge (RRHE) configured to receive the spectrum identification information, optimize the identification parameter based on the received spectrum identification information, and provide the optimized identification parameter to the IoTDs.