Table of Content

30 May 2015, Volume 33 Issue 3

Previous Issue    Next Issue

Communication Engineering

Dynamic Beamforming in Cognitive Radio Networks

GUO Yan, ZHU Fang-jun, LI Ning

2015, 33(3):  223-233.  doi:10.3969/j.issn.0255-8297.2015.03.001

Asbtract ( 1772 )   PDF (5444KB) ( 1755 )  

References | Related Articles | Metrics

Cognitive radio network has the potential of enhancing spectrum utilization.An effective beamforming algorithm can protect primary user from excessive interference while ensuring a meaningful QoS to the secondary system. Conventional beamforming algorithms focus on stationary targets, whereas the cognitive radio nodes are mobile. Dynamic beamforming is investigated in this paper. The direction of arrival (DOA) of a moving target is tracked by a particle filter algorithm. The derived DOA estimation is then used to model the channel between the base station and the users. Three different beamforming
schemes are proposed aiming at maximizing signal-to-noise ratio (SNR), minimizing inference, and SNR balancing, respectively. SNR maximization is transformed to Rayleigh-Ritz quotient, resulting in a closed-form solution. Interference minimization and SNR balancing are reformulated as secondary-order cone programming (SOCP) problems. These problems can be solved efficiently using interior point methods. Several numerical simulation examples
are provided to illustrate effectiveness of our approaches.

Packet Loss Differentiation and Congestion Control in Heterogeneous Networks

JIN Yu-bao, DONG Yu-ning

2015, 33(3):  234-242.  doi:10.3969/j.issn.0255-8297.2015.03.002

Asbtract ( 1775 )   PDF (7714KB) ( 1659 )  

References | Related Articles | Metrics

The existing transmission control mechanism lacks effective method in network
state estimation and packet loss differentiation. To deal with the problem, this paper
proposes a loss differentiation algorithm (LDA), utility based LDA (UBLDA) capable of
estimating the network state (congestion or not) by calculating utility parameters based on
the interval of packet arrival and sending rate. UBLDA can choose appropriate differentiation
scheme based on the state estimation, and integrate with a TCP friendly transmission
control protocol. It is a new multimedia transmission control mechanism suitable for wiredwireless
heterogeneous networks. Simulations are performed to compare the performance
of the proposed algorithm with existing methods.

Dynamic Network Selection Based on Game Theory in Heterogeneous Wireless Networks

XIE Wei1, MA Wen-feng1, HAN Peng2, XU You-yun1

2015, 33(3):  243-252.  doi:10.3969/j.issn.0255-8297.2015.03.003

Asbtract ( 1597 )   PDF (2523KB) ( 1074 )  

References | Related Articles | Metrics

 A main characteristic of the next generation wireless networks is that mobile
terminals can access any networks flexibly. Network selection techniques can improve
spectral efficiency and network capacity. This paper presents an efficient network selection
mechanism based on non-cooperative game. The mechanism can ensure mobile users
to connect to appropriate wireless networks among a heterogeneous network. We show
that the proposed game is a potential game and analyze existence and feasibility of Nash
equilibrium (NE). An iterative algorithm based on best response dynamics for solving the
proposed game is presented. Simulation shows that the algorithm can converge to NE of
the proposed game with a designed utility function, and achieve near optimal performance.

Parallel Optimization of Chinese Language Model Based on Recurrent Neural Network

WANG Long1,2, YANG Jun-an1,2, CHEN Lei1,2, LIN Wei3, LIU Hui1,2

2015, 33(3):  253-261.  doi:10.3969/j.issn.0255-8297.2015.03.004

Asbtract ( 2016 )   PDF (2539KB) ( 2589 )  

References | Related Articles | Metrics

 High computational complexity leads to low efficiency in training a recurrent
neural network (RNN) language model. This becomes a major bottleneck in practical applications.
To deal with this problem, this paper proposes a parallel optimization algorithm
to speed up matrix and vector operations by taking the advantage of GPU’s computational
capability. The optimized network can handle multiple data streams in parallel and train
several sentence samples simultaneously so that the training process is significantly accelerated.
Experimental results show that the model training of RNN is speeded up effectively
without noticeable sacrifice of model performance. The algorithm is verified in an actual
Chinese speech recognition system.

A MAC Layer Broadcast Mechanism in Ad Hoc Networks with Directional Antenna for Transmission and Reception

LI Mo1, HUANG Sheng-bo1, ZHAO Liang2

2015, 33(3):  262-273.  doi:10.3969/j.issn.0255-8297.2015.03.005

Asbtract ( 1770 )   PDF (3359KB) ( 1239 )  

References | Related Articles | Metrics

Using directional antennas for both transmission and reception provides many
benefits for Ad Hoc networks such as longer transmission range and lower probability of
being detected. However, applying traditional broadcast mechanisms for network layers
to Ad Hoc networks will cause long delay problem. To deal with this problem, a broadcast
mechanism is proposed, which uses the reservation sub-frame in the D-TDMA MAC
protocol to reduce the time for delivering broadcast packets. Nodes make reservation for
forwarding broadcast packets after exchanging modified handshaking messages with their
neighbors. This can avoid long delay caused by waiting for receiving broadcast packets
from their parent nodes. Simulation shows that, compared to the performance of planting
traditional network layer broadcast mechanism into an Ad Hoc network, broadcast packets
can be transmitted to all nodes more quickly by occupying smaller slots by using the
proposed mechanism.

Signal and Information Processing

Source Localization Using Taylor Series Iteration with Perturbation Calibration of Emitter States

2015, 33(3):  274-289.  doi:10.3969/j.issn.0255-8297.2015.03.006

Asbtract ( 1807 )   PDF (2911KB) ( 1291 )  

References | Related Articles | Metrics

While most passive localization algorithms are only applicable to certain applications,
the Taylor series iteration (TSI) algorithm is suitable for arbitrary settings. For
this reason, a theoretical framework for deriving localization performance of TSI is discussed.
Two cases, with and without calibration state error, are considered. In the former
case, the corresponding Cramér-Rao lower bound (CRLB) is derived. In the latter case,
the corresponding CRLB is derived and compared with the former case. Mean square error
(MSE) is derived theoretically. Comparison between the theoretical MSE with the CRLB
indicates that accuracy of the source location cannot reach the optimal accuracy without
considering the calibration state error. To overcome the shortcoming, a two-step optimal
fusion localization method based on TSI is proposed, with accuracy reaching that of CRLB.
Passive localization is simulated to verify performance of the proposed algorithm, showing
validity of the theoretical analysis in this paper.  

Adaptive JPEG Steganography Based on New Distortion Cost Function

XU Miao1, CHENG Hang1,2, YU Jiang1, ZHANG Xin-peng1

2015, 33(3):  290-298.  doi:10.3969/j.issn.0255-8297.2015.03.007

Asbtract ( 1609 )   PDF (10035KB) ( 1398 )  

References | Related Articles | Metrics

To improve security of steganography, this paper proposes an adaptive steganographic
scheme for JPEG images based on a new distortion cost function. As texture areas
in a cover image are more suitable for information embedding, the sum of absolute values
of all non-zero quantized DCT coefficients in a DCT block is used to measure texture of
the block. Combining the previous uniform embedding scheme, a new distortion cost function
is defined. With syndrome trellis codes, embedding modification spreads uniformly
over DCT coefficients of all possible magnitudes, embedding secret data adaptively into
the texture areas in the cover. Experimental results show that the proposed scheme out
performs the prior arts in terms of security with the same low payload.

Fuzzy Evaluation of Application Satisfaction for Remote Sensing Satellite

WU Zhao-cong, XU Zhuo-zhi, YANG Fan

2015, 33(3):  299-308.  doi:10.3969/j.issn.0255-8297.2015.03.008

Asbtract ( 1998 )   PDF (1189KB) ( 1613 )  

References | Related Articles | Metrics

Reasonable evaluation of application satisfaction of remote sensing satellites is
crucial for the planning and design of earth observation systems. To deal with this problem,
a fuzzy comprehensive evaluation (FCE) method for remote sensing application satisfaction
is proposed. Based on the remote sensing application of the projects high resolution earth
observation system and space infrastructure planning, a remote sensing application model
is built with four data indicators of space resolution, time resolution, spectral range and
spectral resolution. With the FCE method, a factor set and an evaluation set are designed,
and a series of fuzzy membership function established. By calculating the weight with the
analytical hierarchical process (AHP) method, the final evaluation result is obtained from
a weighted average operator. Comparison between the World Meteorological Organization
(WMO) evaluation cases and the FCE evaluation results shows that using this method to
evaluate remote sensing satellite application satisfaction is feasible.

Super Resolution Reconstruction of ZY-3 Satellite Images

JIA Yong-hong1,2, Lü Zhen1, ZHOU Ming-ting1

2015, 33(3):  309-316.  doi:10.3969/j.issn.0255-8297.2015.03.009

Asbtract ( 1674 )   PDF (2597KB) ( 1468 )  

References | Related Articles | Metrics

 Two refactoring approaches, one based on iterative back projection (IBP) and
the other based on sparse representation and dictionary learning, are discussed. Three
linear array images of the ZY-3 satellite are used to reconstruct super-resolution images.
The reconstruction results are evaluated according to four objective criteria, i.e., mean
brightness, PSNR, information entropy, and sharpness of images. The results obtained with
the two approaches show that the sparse representation and dictionary learning method is
better than the iterative back projection method.

Control and System

Finite Time Stabilization for a Nonlinear Helicopter System

JIANG Yuan, SHI Zhi-qian, DAI Ji-yang

2015, 33(3):  317-328.  doi:10.3969/j.issn.0255-8297.2015.03.010

Asbtract ( 1406 )   PDF (9151KB) ( 964 )  

References | Related Articles | Metrics

A finite-time control method is proposed for the CE150 helicopter with multiple
inputs, strong coupling and nonlinearity. By considering the helicopter system as a cascaded
one and performing coordinate transformation, a finite time nested saturation method is
used to stabilize the two subsystems in finite time. The closed-loop system achieves global
uniform finite-time stabilization by verifying the forward uniformity. Simulation shows
effectiveness of the system, faster convergence, and better robustness.

Robust Estimation of Guaranteed Cost State for Complex Network with Missing Measurements

WAN You-hong, AN Wei-liang, FAN Chun-xia

2015, 33(3):  329-340.  doi:10.3969/j.issn.0255-8297.2015.03.011

Asbtract ( 1636 )   PDF (4326KB) ( 1175 )  

References | Related Articles | Metrics

To deal with data dropouts and noise in data transmission, this paper studies
robust guaranteed cost state estimation of discrete-time complex networks, and focuses on
reducing influence of data dropouts and noises on state estimation. Using the Lyapunov
stability theory combined with stochastic analysis, an existence condition of the gain matrix
and the relevant design criteria used in the robust guaranteed cost state estimation are
proposed with linear matrix inequalities. With convergence of the estimation mean-square
error, the robust guaranteed cost performance requirements are satisfied. Numerical simulation
shows effectiveness and applicability of the proposed design approach.