Research On Network Coding Scheme And Capacity In Wireless Network

The prosperity of mobile data traffic has created tremendous development opportunities to wireless communication networks.Nevertheless,this also imposes higher requirements on network transmission capability,especially the reliable transmission of real-time traffic.As a promising and effective approach,the dense and heterogeneous deployments of wireless communication network are inexorable trend,to get higher spectrum reuse efficiency.However,this change would inevitably imply an increasing complexity of wireless communication environment,and make interference to be the key bottleneck of network performance.Moreover,with the size of wireless network increasing and mobility becoming an inseparable part of networks,the availability of up-to-data channel state information is a challenging task to accomplish or even intractable.Therefore,under this new development situation of wireless communication,it is valuable to investigate wireless network capacity and probe the main constraints of network transmission,to guide the network design.Wireless network capacity,as a key indicator of network transmission capability,which not only gives the limit of reliable transmission capability,but also reveals best ways of transmission,has always been attracting the attentions of researchers.Meanwhile,network coding(NC)has been emerged as a powerful technique to improve the capacity of wireless network.Therefore,to study the effect of interference and Delayed-CSIT on wireless network capacity and probe the limit to timely transmission,this dissertation investigates network coding schemes and capacity from the following two perspectives: 1)In interference network,we study the network coding scheme and explore the capacity of the network,based on delayed channel state information;2)In the context of broadcast packet erasure channel(BPEC),we study the network coding scheme of in-order delivery of the delayconstrained traffic,and analyze its throughput.Concretely,the main contributions of this dissertation are listed as follows:1.In the two-user interference network with permission to utilize past receptions,we propose a network coding scheme based on delayed channel state information,and this scheme can almost achieve the stability region.First,we propose a physical layer linear network coding scheme,to fully use past receptions.Our proposed coding scheme is built upon the following two issues: 1)How to utilize collided packets? 2)How to use the side information? The first issue means that,with feedback and the knowledge of both arrival process,both transmitters should determine how to balance their amount of delivered packets cooperatively.The second issue indicates that,at each transmitter,how to leverage interference by generating the packets desired by both receivers,referred to as packets of common interest.Based on the queue system,we design a feedback-based transmission strategy,to realize the coding scheme optimally.Then,by leveraging the virtual network mechanism,we derive the stability region of the system for the proposed transmission strategy.Meanwhile,a lyapunov-function based stabilizing policy is presented.Finally,we compare the stability region with the capacity outer-bound through extensive simulation.The simulation results reveal that their relative difference is no larger than 3%,implying that the transmission strategy could essentially achieve the outer-bound from the practical perspective.Also,from the comparison with the achievable rate region of some commonly used interference management techniques,the result illustrates that properly exploiting the past receptions can enlarge throughput region.Besides,our work reflects the relationship between stability region and capacity region,in the context of IC with delayed feedback.2.We characterize the capacity region of the two-user interference network with the past reception-based physical layer model.First,we derive the capacity outer bound,by exhibiting that the studied capacity region in this chapter falls within that of the binary interference channel,i.e.,the outer bound for the binary fading interference channel is exactly the desired one.Then,we characterize its linear network coding(LNC)capacity region,which is defined as the capacity region under linear network coding.To facilitate analysis,we classify Linear network codes on the basis of their characteristics,with each subset referred to as coding type.With the definition of coding type,by exploiting a linear-spacebased approach,we unify the problem of probing a capacity outer bound and designing the achievability scheme into a single linear programming(LP)problem.Finally,according to the properties of coding types,by performing pure algebraic arguments,we prove that the LNC capacity region matches the derived outer bound.This implies that the derived LNC capacity region is exactly the true capacity region,and linear coding is optimal.Furthermore,the proving process reveals that utilizing only part of coding types is sufficient for achievability,which enables the simplification of code design.3.For the in-order delivery of timely traffic in broadcast packet erasure channel(BPEC),we design a network coding strategy and analyze its throughput,which demonstrates the existence of network coding gain.Then we derive the throughput of the proposed coding scheme under delay-constraints and no delay-constraints,respectively.To deal with in-order delivery,in the proposed coding scheme,AP only performs coding operation over the oldest native packets in each flow,until the selected packet has been successfully recovered by the associated receiver.Moreover,the coding scheme utilizes the concept of code alignment,which ensures its instantaneous decodability.Then,we analyze the performance of coding scheme under no delay-constraints.By leveraging Markov Decision Process(MDP)to formulate transition process,the throughput of coding scheme can be represented as a set of linear inequalities.For the traffic with arbitrary delay-constraints,queue states cannot be directly used as system states,since the size of corresponding problem is too large to solve.Hence,to reduce the state space,we define lead time.As such,based on the characteristics of coding scheme,by utilizing Markov Decision Process and the system's almost cyclostationary,we obtain the throughput.At last,simulation results reveal that under strict delay-constraints,linear network coding has a better performance than routing,i.e.,there exists coding gain.Meanwhile,the results also illustrate the qualitative impact of strict delay-constraints on the throughput of the system.