Font Size: a A A

Research On Mechanisms For Improving Performance Gain Of Opportunistic Network Coding In Wireless Networks

Posted on:2015-12-04Degree:DoctorType:Dissertation
Country:ChinaCandidate:Q L LiuFull Text:PDF
GTID:1108330473456031Subject:Communication and Information System
Abstract/Summary:PDF Full Text Request
Network coding has been recognized as a promising technique to reduce the number of transmissions, improve the bandwidth utility and thus network throughput and energy efficiency. How to design better network coding architectures and mechanisms to improve its performance gain is a hotspot of academia.The basic idea of network coding is to allow network intermediate nodes to combine several data packets into a coded packet so as to reduce the number of transmissions and improve network throughput. Thanks to the inherent broadcast and snoop features of wireless communications, network coding is especially suitable for the application of wireless environments. In order to facilitate the application of network coding in wireless networks and minimize the modifications for current network protocol stacks, opportunistic network coding is a feasible option. According to the applications of opportunistic network coding in different network environments, we should consider the impact of different factors on the performance gain of opportunistic network coding, such as cache resources, processing ability etc. For different factors, we will build the corresponding mathematic models so as to improve the performance gain of opportunistic network coding. Therefore, for the application of opportunistic network coding in wireless networks, this dissertation identifies the problems that need to be addressed under different network scenarios and situations and proposes opportunistic network coding architectures and mechanisms. The main contents of this dissertation can be summarized as follows.This dissertation first considers the influence of cache resources on the performance gain of opportunistic network coding. The existing opportunistic network coding architectures assume that each node has unlimited cache resources and processing capability. With this assumption, the packets sent or overheard by a network node should be cached in his packet pool for the purpose of possible future decoding. Meanwhile, existing network coding architectures require exchange of information via reception report among neighboring nodes to guarantee the decodability of each coded packet. However, when the cache space of each node is limited, the performance gain of opportunistic network coding will be affected greatly. Even if the resources are infinite, the caching mechanism of packet pool for decoding in existing opportunistic network coding architecture will cause higher network overhead and thus decrease network throughput. To solve this problem, this thesis proposes an opportunistic network coding architecture that optimizes the decoding utility brought by cache resources. Through analysis, we formulate the relationship between cache resources and decoding performance as an optimization mathematical model. By using this model, we derive a series of optimized cache rules and propose a distributed cache policy of opportunistic network coding. Numerical results validate our analytical model and show that our proposed opportunistic network coding architecture can effectively improve the utility of cache resources and thus improve the performance gain of opportunistic network coding.For the application of opportunistic network coding in lossy wireless networks, existing opportunistic network coding architectures rely on pseudo-broadcast to deliver a coded packet to multiple receivers in one transmission and acknowledge the packet receptions by asynchronous ACKs piggybacked in outgoing packets. Within a given pre-set duration, if the sent data packets have not been acknowledged by its intended receivers, they will be inserted in output queue again, coded with other packets and sent out. In lossy wireless networks, this mechanism may easily incur unnecessary retransmissions if asynchronous ACKs are dropped due to packet losses or late arrival and thus waste network bandwidth. In addition, in order to guarantee the reliable delivery of coded packets, existing opportunistic network coding architectures send coded packets by the lowest transmission rate by default, which will cause the longer transmission time and increase the risks of packets collisions. To solve these problems, this dissertation proposes the framework ORC which combines transmission rate control with codes selection so as to achieve the joint optimization of network layer and MAC layer. This framework formulates the transmission rate control of coded packets as a Markov decision process so as to achieve the optimal transmission rate selection. Then, based on the results of rate selection, the optimal packet combination is determined and the resultant coded packet is scheduled for transmission. The packet combination problem can be formulated as the maximum weight clique problem and solved by our proposed code selection algorithm. Network simulations demonstrate that framework ORC can effectively improve network throughput and decrease end-to-end delivery delay.Another major problem of existing opportunistic network coding architectures is the lack of coding opportunities. Due to the randomness of wireless medium access and possible rate mismatch among different codeable flows, there may be insufficient packets to be coded together at the coding node, which compromises the potential throughput gain brought by network coding. Purposely delaying packet transmission can increase coding opportunities. However, this scheme may increase packet delivery delay, which is detrimental for real-time applications. Thus, this dissertation proposes a joint scheduling and network coding framework SNC for real-time traffic flows with diverse delay constraints so as to maximize coding opportunities. Framework SNC includes two mechanisms: Delaying Packet Transmission Policy(DTP) and Network Coding Group(NCG) based Scheduling Policy(GSP). Based on the delay requirements and maximum achievable encoding degree, the former dynamically adjusts the achievable encoding degree so as to maximize coding opportunities. Based on the results of DTP, GSP schedules the coded packets of an NCG for transmission according to the weight so as to improve the throughput gain of existing opportunistic network coding architecture. The network and numerical simulation demonstrate that, framework SNC can maximize coding opportunities so as to improve throughput gain of opportunistic network coding in real-time applications.Although purposely postponing the transmission of packets may increase coding opportunities, existing related works usually focus on some specific network coding structures and have high computational complexity, which is difficult to be realized by a distributed way. Thus, this dissertation proposes a network coding-aware queue management mechanism(DQM), which fully exploits the advantages of network coding in storage compression and further increases coding opportunities. Simulation results demonstrate the effectiveness of DQM.The research achievements of this dissertation provide a theoretical basis and technical means for improving the performance gain of opportunistic network coding and thus improve the overall network throughput and energy efficiency.
Keywords/Search Tags:opportunistic network coding, wireless networks, resource optimization, distributed algorithm
PDF Full Text Request
Related items