Schedulability of Cognitive OFDMA Systems
Final Report Abstract
Dynamic spectrum reuse is an upcoming paradigm for future wireless communication systems. It Is built around the idea that spectrum is no longer explicitly licensed to one system or type of user. Instead, it can be utilized freely by different systems as long as they do not expose other systems, so called primary systems, to too much interference. In this project, we investigated the question if such dynamic spectrum usage impacts the performance of primary systems operating under the most common multi-user communication means of current cellular systems, namely orthogonal frequency division multiplexing (OFDMA). We considered different scheduling principles on top of OFDMA and pursued in general an analytical approach. More precisely, we were interested in evaluating analytically the queuing performance of primary systems under interference patterns typically emerging from dynamic spectrum access. For this, we exploited the well-known effective service capacity notion based on determining the log-moment generating function of the corresponding service processes. Considering the outcome of this analysts we have to highlight the flexibility of the derived models. In contrast to existing literature we provide solutions that allow the free positioning of an arbitrary number of terminals in the cells (no power adjustments such that all terminals have the same signal-to-noise ratio (SNR) / signal-to-interference-and-noise ration (SINR) needed) and a flexible number of subcarriers for each analysis in both noise-limited as well as interference-limited cells. The project established the performance impact in terms of the effective capacity of noise-limited, interference-limited and dynamlc-access-limited scenarios for three schedulers: Round-Robin, proportional fair scheduling and opportunistic scheduling. In case of the strictly interference-limited scenarios, the project found that an exact modelling of the random disturbance by the interfering source is crucial for characterizing the queuing performance. In related work, this impact is often characterized by modelling it as an additional (constant) noise source, which turns out to be a very limited approximation of the real system behaviour. This performance difference is rather small for channel-agnostic schedulers like round-robin (or semi-persistent scheduling in LTE) but is large for channel-dependent schedulers like proportional fair scheduling and opportunistic scheduling. In addition, the system performance in the general interference-limited case is dependent on the absolute averages of the received signal-of-interest and interference signals and not the pure SINR value. While channel-dependent schedulers generally depend more strongly on the average of the signal-of-interest, channel agnostic schemes depend more on the average strength of the interfering signals. Furthermore, the project found for dynamic-access-limited scenarios, that the performance of primary systems with channel dependent scheduling is less drastically impacted by an increasing interference disturbance from secondary systems, in contrast to systems governed by channelagnostic scheduler. This is in particular true for channel-dependent systems operating at lower average strengths of the signal-of-interest, where only marginal performance is lost even if the secondary systems disturb the primary system quite often. In contrast, channel-agnostic schedulers are found to be much more sensitive to the disturbance of secondary interference, regardless of their average signal strength. We believe that the achieved results have important implications for the planning (e.g. cell sizes, influence of femto-/picocells or temporary access points to associated macrocells) and long-term operation (admission control, cell capacity with defined QoS-requirements for each flow) of future cellular networks, in particular if dynamic spectrum access becomes more and more applicable.
Publications
- Effective service capacity analysis of opportunistic multi-carrier OFDMA systems. In Q2SWinet '12: Proceedings of the 8h ACM symposium on QoS and security for wireless and mobile networks, pages 17-26, New York, NY, USA, 2012. ACM
James Gross, Marco Weyres, and Adam Wolisz
(See online at https://doi.org/10.1145/2387218.2387222) - Scheduling with outdated CSI: Effective service capacities of optimistic vs. pessimistic policies. In IEEE 20th International Workshop on Quality of Service (IWOoS), 2012, pages 1-9, 2012
J. Gross
(See online at https://doi.org/10.1109/IWQoS.2012.6245966) - Effective service capacity analysis of interference limited multi-carrier wireless systems. In European Wireless 2013. VDE-Verlag, 2013
Marco Weyres and James Gross
- On the interference as noise approximation In OFDMA/LTE networks. In Proc. of the IEEE International Conference on Communications 2014 (ICC 2014), Melbourne, Australia, June 2014
Donald Parruca and James Gross
(See online at https://doi.org/10.1109/ICC.2014.6883701)
