Research On Cascade Channel Estimation And Performance Evaluation Of Cooperative Communications

Cooperative communication is envisioned as a promising research topic in the communication fields. On one hand, the techniques explore the divergence of the distributed relays, and construct the virtual antenna array. On the other hand, the received signal is amplified and forwarded. All these manipulations improve the quality of the transmitted information. The cascade channel, rather than the individual channel, focuses on the integration of the shadow and fading to the transmission, which easies the comprehension and the application of the cooperative communication. Meanwhile, some cooperative technologies, such as relay beamforming, wireless resource management, relay selection and so on, depend on the knowledge of the cascade channel. Besides, the management and the scheduler of the cooperative communication networks can be more rational based on the statistical properties of the cascade channel. However, the existing estimators of the cascade channel ignore the distribution, which hinders the improvement of the mean square error(MSE). Due to the complicated arrangement, the research and the evalution on the cascade channel remains opening. In this dissertation, the cascade channels of the cooperative communication in the cellular networks and the vehicle-to-vehicle networks are considered. Besides the channel estimation, properties, such as outage probability, average bit error probability(ABEP), ergodic capacity and auto-correlation function, are investigated, respectively.In order to achieve better accuracy of the estimation, a maximum a posteriori(MAP) method is proposed for the cascade channel, based on the Bayesian framework. According to the fact that the cascade channel can be deemed as the product of two independent and Rayleigh distributed ones, the probability density function(PDF) of the cascade channel is derived. The MAP method is deduced based on this PDF and the received trainings. For the purpose of application, the proposed MAP method is simplified by the distribution approximation. The following numerical simulation confirms that the MAP method outperforms the least square(LS) and the maximum likelihood(ML) counterparts in the merit of MSE. It is also illustrated that the simplified MAP method not only remains the estimation accuracy, but reduces the computation remarkably than the original MAP one.In a cooperative communication networks, it is firstly proved that the cascade channel subjects to the Generalized-K distribution when the individual channels between the base and the relays, the relays and the user follow the Nakagami-m and the Rayleigh distributions, respectively. Based on the PDF, the efficiency and the reliability of the cascade channel are both considered with the following merits, such as outage probability, outage capacity, ABEP and the ergodic capacity. Although these merits can be expressed in closed form, it is not convenient for the numerical application owning to the tough Meijer-G function. To this end, the corresponding approximations are derived with the Meijer-G function replaced by the moment-matching(MM) method, the moment generating function(MGF) method, and mixture Gamma(MG) distribution method. The final simulation shows that the MG method fits well with the numerical result better than the other two counterparts, which demonstrates the correctness of the approximation.When the cellular networks with the cooperative techniques are applied in the metropolis or the congested surroundings, the large-scaled shadow deteriorates the transmitted signals remarkably, which needs to be considered. Based on the mentioned cascade channel model, the parameter denotes the power of the channel is modified from a constant to a random variable. Therefore, the small-scaled Generalized-K distributed cascade channel is extended to the composite one which subjects to the Lognormal+Generalized-K distribution. What is more, the corresponding theoretical results and the MG approximated expressions for the outage probability, ABEP, and the ergodic capacity, are also derived. To combat the large-scaled shadow of the composite cascade channel, the diversity technique is considered. Specifically, scenario with a multi-antenna base and many single-antenna relays is discussed both in the optimal selection combining(OSC) and the equal gain combining(EGC) schemes. Although obtaining the theoretical expressions for the merits are intractable, this task is completed in the form of series by the PDF approximation. Simulation results illustrate that the diversity technique improves the efficiency and the reliability notably. Taking the ABEP merit as an example, improvement around 10 d B to 15 d B is achieved with the OSC or the EGC scheme than that without any diversity manipulation.When considering the mobile users and the relays of a vehicle-to-vehicle cooperative communication networks, the temporal correlation is an important merit, which needs to be considered. Due to the Doppler Effect, the above scenario can be portrayed as a geometrical multi-ring and uniform distributed scattering model. The auto-correlation function(ACF) and the Doppler spectrum of the cascade channel can be deduced according to the scattering model. After approximating the time-varying cascade channel as a low-ordered auto-regressive(AR) process, a novel channel estimator based on the Kalman filter is proposed. The convergence of this estimator is also proved. Finally, the numerical simulation exhibits the given estimator achieves better MSE performance than the ML method. More than the temporal ACF and the Doppler spectrum, the other two statistical properties of the mobile cascade channel, which are the level crossing rate(LCR) and the average duration of fading(ADF), are also considered. The expressions of the LCR and the ADF containing single and multiple relays are derived, respectively. Furthermore, according to the different surroundings, both the scenarios with only fast fading and the fast-slow mixed fading are discussed. These two statistical properties are of importance for choosing the frame length for coded systems or the buffer depth for adaptive modulation.