For multi-task mobile edge computing(MEC) systems in satellite Internet of Things(IoT), there are dependencies between different tasks, which need to be collected and jointly offloaded. It is crucial to allocate the computing and communication resources reasonably due to the scarcity of satellite communication and computing resources. To address this issue, we propose a joint multi-task offloading and resource allocation scheme in satellite IoT to improve the offloading efficiency. We first construct a novel resource allocation and task scheduling system in which tasks are collected and decided by multiple unmanned aerial vehicles(UAV) based aerial base stations, the edge computing services are provided by satellites. Furthermore, we investigate the multi-task joint computation offloading problem in the framework. Specifically, we model tasks with dependencies as directed acyclic graphs(DAG), then we propose an attention mechanism and proximal policy optimization (A-PPO) collaborative algorithm to learn the best offloading strategy. The simulation results show that the A-PPO algorithm can converge in 25 steps. Furthermore, the A-PPO algorithm reduces cost by at least 8.87 $\%$ by default compared to several baseline algorithms. In summary, this paper provides a new insight for cost optimization of multi-task MEC system in satellite IoT.

Optical nonlinearity impairments have been a major obstacle for high-speed, long-haul and large-capacity optical transmission. In this paper, we propose a novel convolutional neural network (CNN)-based perturbative nonlinearity compensation approach in which we reconstruct a feature map with two channels that rely on first-order perturbation theory and build a classifier and a regressor as a nonlinear equalizer. We experimentally demonstrate the CNN equalizer in 375 km 120-Gbit/s dual-polarization 64-quadrature-amplitude modulation (64-QAM) coherent optical communication systems. We studied the influence of the dropout value and nonlinear activation function on the convergence of the CNN equalizer. We measured the bit-error-ratio (BER) performance with different launched optical powers. When the channel size is 11, the optimum BER for the CNN classifier is 0.0012 with 1 dBm, and for the CNN regressor, it is 0.0020 with 0 dBm; the BER can be lower than the 7 $\%$ hard decision-forward threshold of 0.0038 from −3 dBm to 3 dBm. When the channel size is 15, the BERs at −4 dBm, 4 dBm and 5 dBm can be lower than 0.0020. The network complexity is also analyzed in this paper. Compared with perturbative nonlinearity compensation using a fully connected neural network (2392-64-64), we can verify that the time complexity is reduced by about 25 $\%$ , while the space complexity is reduced by about 50 $\%$ .

A compact all-fiber magnetic field sensor based on the magnetic gel coated hollow core fiber (HCF) has been proposed and experimentally demonstrated. A double-layered Fabry–Perot resonator is formed by coating a magnetic gel layer on the surface of the HCF. Anti-resonant reflecting guidance of light can be achieved in the HCF, and leaky mode is induced at resonant wavelength of the double-layered Fabry–Perot resonator, which results in lossy dips in the transmission spectrum of the HCF. Due to the tunable magneto-elastic effect, the shape of the magnetic gel is deformed with the external magnetic field, which results in a change of the resonate condition for the double-layered Fabry–Perot resonator. The magnetic field can be measured by interrogating the wavelength of the lossy dip. The experimental results show that a magnetic field sensitivity of $245\mathrm{pm}/\mathrm{Oe}$ is achieved.

A capillary covered silica hollow core fiber (HCF) has been designed and tested for the measurement of displacement based on antiresonant reflecting optical waveguide. A section of the silica HCF was inserted into a silver coated capillary. A Fabry-Perot resonator can be formed in the silica cladding. The leaky mode of the guided light can be achieved at resonant wavelengths of the Fabry-Perot resonator, which results in lossy dips in the transmission spectrum. The transmission power of the dip is sensitive with the displacement of the capillary since the effective reflectivity of the Fabry-Perot resonator is affected by the location between the capillary and the silica HCF. The experimental results show that the sensitivity of up to 0.578 dB/μm is achieved, and the proposed sensor is insensitive with the temperature.

The strong stochastic nonlinear impairment induced by random mode coupling appears to be a long-standing performance-limiting problem in the orbital angular momentum (OAM) mode division multiplexing (MDM) of intensity modulation direct detection (IM/DD) transmission systems. In this Letter, we propose a Bayesian neural network (BNN) nonlinear equalizer for an OAM-MDM IM/DD transmission with three modes. Unlike conventional Volterra and convolutional neural network (CNN) equalizers with fixed weight coefficients, the weights and biases of the BNN nonlinear equalizer are regarded as probability distributions, which can accurately match the stochastic nonlinear model of the OAM-MDM. The BNN nonlinear equalizer is capable of adaptively updating its weights and biases sample-by-sample, according to the probability distribution. An experiment was conducted on a 300-Gbit/s PAM8 signal with three modes over a 2.6-km OAM-MDM RCF transmission. The experimental results demonstrate that the proposed BNN nonlinear equalizer exhibits promising solutions to effectively mitigate nonlinear distortions, which outperforms conventional Volterra and CNN equalizers with receiver sensitivity improvements of 1.0 dBm and 2.5 dBm, respectively, under hard-decision forward error correction (HD-FEC) thresholds. Moreover, compared with the Volterra and CNN equalizers, the complexity of the OAM-MDM is significantly improved through the BNN nonlinear equalizer. The proposed BNN nonlinear equalizer is a promising candidate for the high capacity inter-data center interconnects.

With the rapid development of next-generation mobile network services, there is a growing need for customized services to meet the demands of various network functions. Leveraging the Software-Defined Networking (SDN) architecture, Network Function Virtualization (NFV) enhances service delivery flexibility by virtualizing network appliances. This allows for Service Function Chain (SFC), which further enhances service delivery flexibility through centralized, programmable management. However, existing works require manual adjustments and tuning when adapting to evolving user demands and network expansions, lacking the flexibility needed for changing network conditions. With the rise of Large Language Models (LLMs), the automation of network management has gained new momentum by understanding programming logic, generating code, and incorporating advanced knowledge of network and optimization. This paper introduces an LLM-assisted network operating system framework and presents a case for LLM-assisted SFC optimization. Finally, it proposes an NSGA2-based multi-objective LLM optimization algorithm, which continuously updates the heuristic code policies through evolutionary iterations. Simulation results validate the effectiveness of this approach in achieving stable and efficient multi-objective optimization for SFC deployment.