Convolutional neural network based network traffic classification scheme suffers many disadvantages such as the complex structure designing, gradient declines or even explodes, the deterioration of prediction accuracy, and etc. A residual network based improved traffic classification algorithm is proposed. The convolution layer and pooling layer in the traditional convolutional neural network are replaced by the residual network layer, which can alleviate the problem that the traditional convolution network is too deep to train effectively. The data feature information learned by the proposed algorithm in the training stage is more comprehensive, and the trained model can also be more accurate. Simulation results show that the improved algorithm has higher accuracy than the traditional neural network, which can be improved from 92.05% to 96.18%.

In order to improve the performance of indoor Visible Light Communication(VLC) system, a hybrid modulation scheme combining Carrierless Amplitude Phase(CAP) modulation and Digital Pulse Interval Modulation (DPIM) is proposed. This scheme takes advantage of both characteristics of higher spectral efficiency and power efficiency. A parallel modulation structure is adopted to realize the signal generation at the transmitter side, and the transmission of Direct Current biased Optical-CAP(DCO-CAP) signal are controlled by pulse time slot of the pulse interval modulation. The threshold value detection and the maximum likelihood criterion are used to complete the serial demodulation of modulated signals at the receiver side. Analytical expressions of Bit Error Rate (BER) and Frame Error Rate (FER) for the hybrid modulation scheme over Gaussian channel are derived. In addition, the performance of the proposed scheme and ordinary schemes are contrastively analyzed in terms of spectral efficiency and system reliability. The numerical results show that the hybrid system is more reliable compared with ordinary modulation system under the condition of same spectral efficiency.

The basic requirement of the 5th Generation (5G) network is to realize automatic end-to-end service arrangement and slice orchestration. In this paper, considering the purpose of the open standardization of the Northbound Interface (NBI) of 5G transport network, we first analyze the functional requirements of the NBI. Then the modeling scheme of the NBI is proposed. Finally, the different technical implementation schemes of 5G transport network are shielded, which provides a reference for the unification of the NBI.

Under the background of vigorously developing smart grids, the scale of power optical communication networks supporting power grid operation is becoming larger and larger, and the services carried are more diversified. However, the service routing planning of the power communication network is mainly based on the shortest path algorithm, which leads to the imbalance of the business importance distribution of the power communication networks. Therefore, the local risk of the network can be high and the overall health of the network is low. Aiming at the shortcomings of traditional routing algorithms, this paper proposes a route algorithm that uses deep reinforcement learning technology to balance the risk of network traffic, which also comprehensively considers the traditional constraints such as optical transmission constraints and link residual capacity. The algorithm considers the distribution of service importance, link capacity, and link optical signal-to-noise ratio to achieve risk equalization of power communication networks. We have carried out an experiment in a provincial power communication subnet. The result shows that the method can effectively reduce the risk balance of the power optical communication networks and provide a strong guarantee for the safe operation.

The development of city agent puts forward new requirements for network system. As an important branch of city agent, public security agent also has a further demand for the network, which needs the network to be more active, fast and flexible to support its development. Another requirement of the network is to be more intelligent. Considering these issues, this paper investigates the Artificial Intelligence (AI) network architecture based on public security agent. We also propose a platform framework of public security agent and the AI network architecture based on public security agent. Finally, some application scenarios in the field of public security are given.

Aiming at the problem that the number of single disks of traditional Optical Transport Network (OTN)equipment business is too small, and the expansion of the slot will cause business interruption and high cost, this paper proposes a Cluster Architecture of Optical Transport Network (COTN) communication equipment. The design of the COTN equipment management system is an effective technical guarantee to upgrade existing network equipment from OTN to COTN, which is the only technical means to ensure the compatibility and scalability of the single-disk service bearer in COTN. Therefore, this paper adopts a single network element and multiple sub-frames. A comparative study of COTN equipment and traditional OTN equipment in the areas of data collection, inter-frame protection, and service processing show that the expansion of COTN equipment service slots meets the low-cost and dynamic sustainability requirements of bearer service expansion.

Ethernet transmission has high requirements of clock synchronization accuracy. Factors such as switching between different clock domains and data bit width conversion will enlarge the jitter amplitude of the time stamp in the Precision Time Protocol (PTP) frame, resulting in the problem of clock asynchronization. In order to solve this issue, this paper proposes a method that can eliminate the jitter of the time stamp to improve the accuracy of the time stamp-uniform envelope rate characterization, which ensures the uniformity of the Ethernet transmission data distribution by depicting a uniform data envelope. Therefore, the time stamp jitter can be eliminated and precise clock synchronization is achieved. Experimental data shows that after the data is characterized by a uniform rate envelope, the jitter amplitude of the time stamp is significantly reduced with obvious the jitter elimination, and the jitter range remains within ±20 ns.

In order to extend the transmission distance of the traditional C-band high-speed optical module, this paper studies a kind of quasi-coherent reception technology based on the existing 25 Gbit/s optical module, and discusses the application scheme of the quasi-coherent reception technology. It is found that the quasi-coherent receiver can effectively reduce the influence of dispersion on the transmission distance of high-speed optical module without using additional electrical or optical dispersion compensation technology. The experimental results show that the quasi-coherent receiver has strong processing ability for optical signals degraded by dispersion, and the use of the quasi-coherent receiver can significantly improve the transmission sensitivity of the optical module. The experimental results show that the quasi-coherent reception technology can extend the transmission distance of the C-band 25 Gbit/s optical module with traditional intensity modulation technology from 10 km to more than 25 km.

Spatial division multiplexing based on the multi-core fiber has been believed as a promising and effective way to overcome the foreseeable capacity crunch of single mode fiber transmission. Fan-in Fan-out (FIFO) technology for multi-core fiber is the key to realize multi-core fiber space division multiplexing system, which can realize the optical coupling function from single multi-core fiber to multiple single core fiber. In this paper, we summarize the methods of multi-core fiber FIFO devices in the past decade, which includes fused type, fiber bundle type, free space type and three-dimensional integrated waveguide type. The fabrication methods and performance of several types of devices are analyzed and compared. Finally, the technology direction which can be widely used in the future is prospected.

In order to solve the problem that the pin of butterfly laser connected with lead wire breaks in the sweep frequency vibration test, the finite element analysis model is established with different pin length to reflect the influence of lead wire. The stress of the pin is analyzed by mode superposition method under the condition of 20 g and 20 ~ 2 000 Hz sinusoidal sweep frequency vibration. Based on Miner criterion, the life of pin under vibration is calculated. The analysis results show that the natural frequency of the original laser pin is higher than 2 000 Hz. After connecting to the lead wire, the natural frequency will be reduced to less than 2 000 Hz, and the resonance will occur under the vibration condition. It is also shown that if the lead wire is fixed appropriately, the natural frequency can be increased to more than 2 000 Hz to avoid resonance. Moreover, the results indicate that the original laser pin has infinite life under vibration condition, but the life of the pin connected by lead wire is greatly reduced. The pin life can be increased to infinite life by fixing the lead wire appropriately.

To deal with issue of high hardware cost and energy consumption, beam selection is used in the lens antenna array based millimeter-Wave (mmWave) massive Multiple-Input Multiple-Output (MIMO) system to reduce the required number of radio frequency chains. However, beam selection requires the base station to acquire the accurate channel state information. To solve this problem, we first use the structural characteristics of beamspace and adopt a Support Detection (SD)-based channel estimation scheme to estimate the large channel with lower pilot overhead and computational complexity. Then, inspired by Ant Colony Optimization (ACO) algorithm in bionics, an ACO beam selection scheme based on SD channel estimation is proposed. The scheme can not only avoid the interference of beams between users, but also maximize the system sum rate. Finally, the simulation results show that the proposed scheme can obtain a near-optimal solution with significantly lower computational complexity, and it is superior to the existing schemes in improving the system sum rate.

Lognormal channel is a powerful mathematical tool in modeling fading effects for wireless optical communications over weak turbulence channels. However, exact analysis involving lognormal channels involves multi-fold integrals, which reveal little insights into the relation between the system performance and the channel parameters. Therefore, Wilkinson Approximation (WA) and Fenton Approximation (FA) are widely applied to approximate the performance of lognormal fading channels. However, their ranges of applications are not fully studied, which causes discrepancies in performance evaluations. In this work, we compare WA and FA based on the expressions of moments, and propose an approach to approximate the performance of Free-Space Optical Communications (FSO) relaying systems over the lognormal weak turbulence channels, and discuss the ranges of applications for WA and FA. It is shown that at low and medium Signal-to-Noise Ratio (SNR), error probability approximation based on WA is more accurate. However, at large SNR, Fenton’s approach will provide better approximation accuracy compared to the WA. Simulation results indicate that the approximation error depends on the discrepancies between the variances over FSO links as well as the scintillation level of the channels.

Terminal equipment represented by inspection robots is widely used in substations. In the future, 5th-Generation (5G) mobile terminals will be more and more used in substations with the fast development of 5G technology. The performance of the communication system is closely related to the wireless channel. Therefore, it is very important to study the 5G communication channel characteristics of the mobile terminal in the substation. Aiming at the characteristics of the communication channel of 5G mobile terminals in the Three-Dimension (3D) scattering environment of substations, based on the multiple input multiple output technology, the article proposes to use geometric analysis to establish a 3D channel model, and derive the time autocorrelation function and spatial cross correlation function of the channel. Based on the 5G frequency band, the auto-correlation and cross-correlation characteristics of the wireless channel are simulated and analyzed. The influence of the Rice factor of different sizes on the wireless channel characteristics is also studied. The above simulation results show the availability of 5G terminals in substations, and broaden the research on mobile terminals using 5G communication technology in substation scenarios.

In order to generate high-frequency millimeter-wave, this paper proposes a scheme of generating 16-tupling millimeter wave signal with 4 polarization modulators in parallel, using the different polarities of optical sideband to merely reserve optical carrier and the 8nth-order sidebands. By properly selecting the modulation index, we can get the ±8th-order optical sideband signal with carrier suppressed. Finally, after the composite optical signals beaten on the photodetector, the 16-tupling millimeter-wave signal is achieved. The principle of generating 16-tupling millimeter-wave signal is theoretically analyzed. The optical sideband suppression ratio of the obtained ± 8th-order sideband signal is 61.72 dB, and the radio frequency stray suppression ratio of the 16-tupling millimeter-wave signal is 55.70 dB. The optical sideband suppression ratio and the radio frequency stray suppression ratio obtained from the simulation are 60 dB and 53 dB, respectively. The effects of phase offset and modulation index on sidebands suppression ratio are also analyzed. The experimental results are consistent with the theoretical analysis, which can verify the feasibility of the proposed scheme.

To meet the demand of 5G mobile base stations and edge computing nodes for high-precision ground timing system, the construction method of end-to-end ground time synchronization links and the subsequent time synchronization network planning ideas are proposed, which do not rely on satellite timing system. The feasibility of the proposed method is verified by conducting evaluation experiments in the present 5G mobile network of SuZhou, which demonstrate the high and stable time accuracy within 24.8 ns for 72 hours. Key issues such as satellite independent time synchronization network are also analyzed to point out the future improvement direction.