Authors: Binghan Wu, Kai Zeng, Zhiqi Li, Zhigang Wang, Yang Liu, Ya-Qin Zhang, Joseph Sifakis, Ye Ouyang
Organizations: AsiaInfo Technologies, Beijing, China · Hong Kong Polytechnic University, Hong Kong, China · Institute for AI Industry Research (AIR), Tsinghua University, Beijing, China · Verimag, Université Grenoble Alpes, Grenoble, France
Learned models for radio access networks (RANs) are often developed using simulation, yet their predictions can shift when exposed to physical propagation conditions, protocol behavior, and equipment implementations. This paper presents a cyber-physical recursive self-improvement (RSI) framework for AI-RAN that connects a task-oriented student model, an experiment-planning coach agent, a calibratable virtual environment, and a physical host agent. Physical measurements are organized as evidence for both model adaptation and subsequent experiment planning. We examine an implementation of the physical-feedback pathway using ping-pong handover as a case study. The system completes 12 physical executions organized into six baseline--candidate pairs, updates the student from measured outcomes, and evaluates the revised model on reserved physical configurations. In the observed short-window tests, increasing the time-to-trigger from 320 to 1280,ms changes the ping-pong count from three to zero. The updated student achieves a grouped-validation mean absolute error (MAE) of 2.673 round trips/min, while a common three-configuration physical evaluation yields MAEs of 1.240 and 1.601 round trips/min for the pre-feedback and post-feedback models, respectively. These results characterize both the operational feedback loop and the parameter-dependent generalization behavior revealed by new physical experiments.
Figures & tables
Figure 1: Cyber-Physical RSI architecture. Solid arrows indicate the implemented physical-experiment and student-feedback pathway; dashed arrows show the virtual-environment alignment and candidate-screening extension. Experimental evidence also enters the coach’s versioned memory for subsequent planning.
Block
Varied parameter
Candidate
Nbase
Ncand
rbase
rcand
Learning
A3 offset
2 dB
2
3
7.130
10.696
Learning
Hysteresis
2 dB
3
3
10.680
10.696
Learning
TTT
512 ms
3
1
10.692
3.563
Evaluation
A3 offset
3 dB
2
3
7.122
10.685
Evaluation
Hysteresis
3 dB
2
3
7.116
10.686
Evaluation
TTT
1280 ms
3
0
10.677
0.000
Table 1: Physical ping-pong response in six baseline–candidate pairs. The common baseline is A3 offset =1 dB, hysteresis =1 dB, and TTT =320 ms. Frequencies are in round trips/min.
Figure 2: Observed ping-pong frequency for the three evaluation candidates and their independently executed paired baselines. Each bar represents one scoring window of approximately 16.8 s.
Model
Dataset
Event Brier ↓
Effect MAE ↓
V2
Before feedback
0.0000806
3.1942
V3
After feedback
0.0000819
2.8839
V4
After feedback
0.0000819
3.5832
V5
After feedback
0.0000819
3.6369
V6
After feedback
0.0000819
2.6730
Table 2: Session-grouped validation of student candidates. V2 uses the pre-feedback dataset, whereas V3–V6 use the updated dataset. The effect-prediction MAE is in round trips/min.
Candidate
Observed
V2 pred.
V6 pred.
V2 ∣e∣
V6 ∣e∣
Offset =3 dB
10.6848
11.0651
11.7037
0.3804
1.0189
Hyst. =3 dB
10.6859
11.0594
11.6997
0.3735
1.0138
TTT =1280 ms
0.0000
2.9664
2.7715
2.9664
2.7715
MAE
—
—
—
1.2401
1.6014
RMSE
—
—
—
1.7401
1.8025
Table 3: Predicted and measured ping-pong frequencies for the common reserved physical evaluation block (round trips/min). MAE and RMSE give equal weight to each candidate.
Figure 3: Absolute configuration-effect prediction error before feedback (V2) and after feedback (V6), evaluated against the same reserved physical measurements.
To make cross-band channel prediction practical for AI-native RAN, algorithms must generalize across diverse environments and support real-time inference. Existing approaches achieve one but not both. To bridge this gap, we introduce GUIDE, a physics-guided deep unfolding framework that embeds wireless channel physics into differentiable layers. Without retraining in unseen environments, GUIDE achieves 2.75x beamforming gain than the deep learning-based baseline FIRE with only a slight increase in inference time, and 1.39x beamforming gain than the strongest model-based baseline R2F2 while running over 1610x faster.
Ruiqi Kong, He Chen, Xiaojun Lin
Department of Information Engineering, The Chinese University of Hong Kong, Hong Kong SAR, China
Recursive self-improvement (RSI) describes a system that improves not only its behavior but also the process by which it improves. Recent digital agents test this idea by rewriting their own code and checking each version with executable tests. Bringing these mechanisms into the physical world requires learning from partial observations through costly trials, while accounting for changing operating conditions and potentially irreversible consequences. We position RSI within robot learning and trace pathways from improving robot capabilities to improving the processes that generate them. We examine how physical experience can strengthen these improvement mechanisms and when the resulting gains can support further self-improvement of the mechanisms themselves. Gains carry forward only when feedback addresses remaining learning needs, the receiving component can use the change, its value survives changes in tasks or hardware, and continued practice remains safe and affordable. Existing studies, concentrated in manipulation, support behavioral improvement, benefits to later learning, and procedure improvement. Some systems also change the improver's knowledge or tools, but the physical studies reviewed here have yet to isolate gains attributable to recursion. We synthesize these findings into evaluation principles and research directions for machines that, trial by trial, safely improve the way they improve.
Meng Luo, Zechen Bai, Shanqing Xu +9
National University of Singapore · Huazhong University of Science and Technology · Shanghai Jiao Tong University +2
Cellular research and development (R&D) is throttled by six structural processes that each consume months of manual engineering work per iteration: (i) synthesizing new features from standards or research papers into production code; (ii) conformance and interoperability testing; (iii) hardening against field anomalies and diverse deployment environments; (iv) data-driven optimization of network functionalities; (v) discovering and prototyping novel waveforms, functionalities, and capabilities for future standards; and (vi) securing the stack against vulnerabilities. Although Large Language Models (LLMs) have compressed comparable R&D work in general software engineering from days to minutes, their known pitfalls worsen on Radio Access Network (RAN) use cases: they hallucinate Application Programming Interfaces (APIs) and mis-read specifications, which kills interoperability of RAN components at the first mistake, and they heavily rely on simulations for designing algorithms, which is notorious for breaking when transferred to real hardware. To address these challenges, we present GENESIS, an agentic Artificial Intelligence (AI) framework that converts intents (e.g., a specification clause, a telemetry anomaly, or a research hypothesis) into solutions validated with over-the-air experiments, fed back into a persistent knowledge base. GENESIS is built on three composable primitives (agents, skills, hooks) and a knowledge layer (SYNAPSE) that doubles as the source of ground truth and the recipient of every artifact the framework produces, making capabilities compound across runs.
Tamerlan Aghayev, Maxime Elkael, Michele Polese +11
Institute for Intelligent Networked Systems at Northeastern University, Boston, MA.