Instantaneous and ergodic rates are two of the most commonly used metrics to characterize throughput of wireless networks. Roughly speaking, the former characterizes the rate achievable in a given time slot, whereas the latter is useful in characterizing average rate achievable over a long time period. Clearly, the reality often lies somewhere in between these two extremes. Consequently, in this work, we define and characterize a more realistic N-slot average rate (achievable rate averaged over N time slots). This N-slot average rate metric refines the popular notion of ergodic rate, which is defined under the assumption that a user experiences a complete ensemble of channel and interference conditions in the current session (not always realistic, especially for short-lived sessions).
The proposed metric is used to study the performance of typical nodes in both ad hoc and downlink cellular networks. The ad hoc network is modeled as a Poisson bipolar network with a fixed distance between each transmitter and its intended receiver. The cellular network is also modeled as a homogeneous Poisson point process. For both these setups, we use tools from stochastic geometry to derive the distribution of N-slot average rate in the following three cases: (i) rate across N time slots is completely correlated, (ii) rate across N time slots is independent and identically distributed, and (iii) rate across N time slots is partially correlated. While the reality is close to third case, the exact characterization of the first two extreme cases exposes certain important design insights. / Master of Science / Choice of an appropriate metric is essential for accurate design and analysis of wireless networks. The two most popular metrics used to characterize data rate or throughput of wireless networks are instantaneous and ergodic rates. While instantaneous rate characterizes the throughput achievable in a given time slot, the ergodic rate characterizes the average achievable throughput over a long period of time. But often, the real-world scenarios fall in between these two extremes, where the network performance is to be characterized over a given finite number of time slots. Hence, we define and characterize a new suitable metric <i>N-slot average rate</i>, which is the achievable rate averaged over <i>N</i> time slots.
Using this metric, we develop an analytical framework to study the performance of typical nodes in both ad hoc and downlink cellular networks. We model these networks using homogeneous Poisson point processes and characterize the <i>N</i>-slot average throughput using the tools of stochastic geometry. Accounting for the prominent cases of network mobility, we derive the distribution of <i>N</i>-slot average rate in the following three scenarios: rate is completely correlated across <i>N</i> time slots, rate is independent and identically distributed across N time slots, and rate is partially correlated across <i>N</i> time slots. We studied the impact of various system parameters on our metric and also discussed key insights from our results.
| Identifer | oai:union.ndltd.org:VTETD/oai:vtechworks.lib.vt.edu:10919/85104 |
| Date | 30 March 2017 |
| Creators | Bodepudi, Sai Nisanth |
| Contributors | Electrical and Computer Engineering, MacKenzie, Allen B., Dhillon, Harpreet Singh, Reed, Jeffrey H. |
| Publisher | Virginia Tech |
| Source Sets | Virginia Tech Theses and Dissertation |
| Detected Language | English |
| Type | Thesis |
| Format | ETD, application/pdf |
| Rights | In Copyright, http://rightsstatements.org/vocab/InC/1.0/ |
Page generated in 0.0022 seconds