Study of steady-states in distributed data caching in ad hoc networks
Abstract
There has been extensive research on cooperative distributed data caching in ad hoc networks.
However, most of the work has focused on how to reduce the average delay of requests and
improve the packet delivery ratio, etc; not much work has been done to study the steady-state
status achieved by distributed caching algorithms. Information related to steady-state status
includes the convergence time of the caching algorithms, the final data cache placement in the
network, the stabilized cost performance, and the performance comparison of distributed caching
algorithms with an optimal centralized caching solution. Previous theoretical results show that to
minimize the average access cost in the network, the optimal number of replicas of each data
object is proportional to the square root (or two-thirds) of the data’s access frequency. In this
work, we empirically show that the optimal replica number not only depends on the access
frequencies of data, but also depends on the storage capacity of each node. We propose a
heuristic model studying both cooperative, hybrid, and selfish caching steady-states in ad hoc
networks. We formulate and solve the data caching problem optimally using integer linear
programming (ILP) in order to validate our findings with regard to access frequency. We also
provide empirical data regarding the steady-state cost of data based on the storage capacity of the
nodes in the network. Via extensive ns-2 simulations [10], we gain some insight regarding the
steady-states of distributed data caching.
Description
Thesis (M.S.)--Wichita State University, College of Engineering, Dept. of Electrical Engineering and Computer Science