报告详情
Link-Aware Hybrid PSO–DBO Routing and Communication-Efficient Federated Crop Recommendation in Agricultural Sensor Networks
编号：30
访问权限：仅限参会人
更新：2026-11-10 16:14:50 浏览：8次
Online
报告开始：2026年11月20日 17:15（Asia/Kolkata）
报告时间：15min
所在会场：[S8] Signal Processing, RF and Communication Hardware/6G and Future Wireless Communication Systems [S8-1] Signal Processing, RF and Communication Hardware/6G and Future Wireless Communication Systems
暂无文件
摘要
Agricultural wireless sensor networks (AWSNs) must deliver field data on a few joules per node, and they increasingly feed learning-based services such as crop recommendation. Existing work optimizes routing as if only sensing traffic existed, and it trains recommenders centrally on raw farm records. We present HOFL-AWSN, which couples the two. A hybrid optimizer (HPDO) combines particle-swarm search, dung-beetle thief re-seeding and a memetic cluster-head-swap local search. It selects cluster heads and relay trees under an objective that includes link quality, predicted energy, delay, cluster balance and the expected federated-learning (FL) load. Farm gateways train a crop recommender with EA-FedPC, an FL scheme that uploads top-k
, 8-bit, error-feedback updates and aggregates them with server momentum. In a co-simulation over 10 seeds on the public Crop Recommendation dataset, HOFL-AWSN raises the packet delivery ratio of a 200-node network from 64.02% to 91.74% and its energy efficiency by 52.6% relative to a re-implementation of FDNO. In a 100-node network, delivery, delay and last-node lifetime improve, but the first and half of the nodes die earlier. EA-FedPC reaches 97.11% accuracy under label skew (centralized: 98.48%), matching momentum FedAvg with 12× fewer uplink packets.
, 8-bit, error-feedback updates and aggregates them with server momentum. In a co-simulation over 10 seeds on the public Crop Recommendation dataset, HOFL-AWSN raises the packet delivery ratio of a 200-node network from 64.02% to 91.74% and its energy efficiency by 52.6% relative to a re-implementation of FDNO. In a 100-node network, delivery, delay and last-node lifetime improve, but the first and half of the nodes die earlier. EA-FedPC reaches 97.11% accuracy under label skew (centralized: 98.48%), matching momentum FedAvg with 12× fewer uplink packets.关键词
Agricultural WSN, energy-efficient routing, hybrid metaheuristic, federated learning, crop recommendation, non-IID data
报告人
Yakkala Lavanya
RESEARCH SCHOLAR JNTUA稿件作者
全部评论
重要日期
-
会议日期
11月20日
2026
至11月22日
2026
-
10月31日 2026
初稿截稿日期
-
11月07日 2026
提前注册日期
-
11月22日 2026
注册截止日期
主办单位
United Societies of Science
承办单位
Pondicherry University
协办单位
IEEE Section
IEEE Madras Section
IEEE Madras Section
联系方式
- ICon INDIA 2026
- ic******@usssociety.org
发表评论