Study Of Data Disseminating Schemes Based On Data Scheduling And Retrieval In Wireless Networks

Wireless data broadcast is an effective way for data dissemination. In wireless networks, because of the increasing number of mobile users and the limitation of broadcasting infrastructure, it is important to efficiently disseminate a variety of information to a large number of mobile users. Thus, significant research interest has been drawn to data scheduling and retrieval to achieve different goals, such as quick data dissemination, prolonging the battery life time of mobile devices, improving the scalability of networks, utilizing the bandwidth and saving the waiting time of users.In the research of data scheduling, we first study how to place data items at multiple channels for solving different problems, such as bandwidth scarcity of channels and the limited battery life of mobile devices. According to different data sources, data scheduling is studied from two aspects:based-push and based-pull wireless data broadcast (on-demand data broadcast). The advantage of the former is that the preparation time of broadcasting data items at servers is less, and the disadvantage is that the order of data items are fixed during the broadcasting process of data items such that the requirements of all mobile users may not be satisfied. While the advantages of the latter are that the broadcasting order of data items can be adjusted during the broadcasting process to satisfy the requirements of all mobile users, and the disadvantages are that the preparing time of broadcasting data items at servers is more and it is difficult to hold the best performance. In the research of data retrieval, we first study how to find a data access pattern among multiple channels, so that the waiting time and energy consumption are reduced. According to the number of equipped to mobile users, data retrieval is studied from two aspects:single-antenna and multi-antennae. The former has the following drawbacks of ignoring the conflicts of data retrievals in two different channels in adjacent time slots (2-slot conflicts) and the time constraints for data download from user requests, and is applicable only for single request data retrieval. The latter does not consider the relationship among data items of different requests for multi-request data retrieval. Corresponding to the above drawbacks, the main contributions of this dissertation are as follows:(1) Deadline-constrained on-demand data broadcast for avoiding conflicts. For avoiding two conflicts of data scheduling and reducing the number of channel switching, we design a strategy under the deadline constraint in the request:if the deadline allows, we place all requested data items at the channels; otherwise, we place the maximum number of requested data items allowed by the deadline to reduce data miss ratio. The experimental results show that the proposed schemes can efficiently solve on-demand data scheduling problem, attain the reasonable data broadcast sequences, shorten the broadcast cycle and improve the performance of data dissemination.(2) On-demand multimedia data broadcast. We propose several schemes to solve this problem:the first scheme applies a bipartite graph to hold the most suitable sequence for placement according to the broadcasting time of multimedia data we; the second scheme divides the longest multimedia data item to further balance the broadcasting cycle of all channels. The experimental results show that the proposed schemes can further reduce the difference of broadcast cycles among the channels.(3) Energy-efficient cooperative MIMO routing in wireless senor networks. We focus on the energy consumption problem of sensor nodes. To prolong the life time of WSNs, we minimize the energy consumption of disseminated data by sensor nodes. The algorithm introduces the idea of variable cluster size for balancing between the residual energy of cluster head and cluster size. The experiment results show that the proposed schemes can balance the energy consumption of each cluster, and reduce the total energy consumption of sensor wireless networks.(4) Non-consecutive data parallel retrieval. We propose a parallel data retrieval algorithm for solving the non-consecutive data parallel retrieval problem in multi-antennae MIMO wireless broadcast networks, such that the access latency and energy consumption are reduced. In this algorithm, we compute the weight of channels by two balance factors and select the suitable retrieval channel. The experiment results show that the proposed schemes can achieve the parallel data retrieval.(5) Approximate data retrieval for avoiding 2-slot conflicts. We convert the data retrieval problem under 2-slot conflicts to set cover problem by constructing a DAG for the cases of both single antenna and multi-antennae. Both can be solved by directly applying the greedy set cover scheme, while the latter needs to firstly divide the requested data items. The experiment results show that the proposed algorithms have the characteristics of higher accuracy and lower time complexity. Comparing with the existing algorithms, they have advantages.(6) Multi-antennae multi-requests data retrieval. For multi-requests data retrieval, we convert the wireless data broadcast system to a super tree, such that the number of all possible data retrieval sequences is smaller than that of a DAG, which reduces the time of computing the set cover. For multi-antennae data retrieval, we propose two different grouping schemes to allocate the requests to different antennae. The experiment results show that the proposed schemes can balance the access latencies among the antennae. Comparing with the existing algorithms, they have advantages.