Datasets
Standard Dataset
Hypergraph-based Modelling for Coexistence-Aware Channel Allocation
- Citation Author(s):
- Submitted by:
- Tawachi Nyasulu
- Last updated:
- Fri, 01/26/2024 - 10:12
- DOI:
- 10.21227/ybr7-1728
- Data Format:
- License:
- Categories:
- Keywords:
Abstract
In this paper a novel technique for modelling a radio frequency (RF) environment based on hypergraph theory is investigated for solving coexistence management of heterogeneous networks and efficient channel allocation for spectrum sharing. Conventionally, traditional graph theory is used to model interference relationships and exclusive channel allocation. The demand for wireless services is increasing, hence the need for efficient spectrum management techniques, such as spectrum sharing among coexistent networks. However, coexistence-aware channel allocation would require representation of both interference and spectral coexistence relationships in the RF environment model. A graph cannot be used to represent multiple and multifaceted relationships without violating consistency with graph theory. This paper therefore proposes representing the RF environment using a hypergraph. The network coexistence method used is an implementation of the IEEE 802.19.1 method for co-sharing via network geometry classification. The simulation results show that the hypergraph-based model allocated channels, on average, up to 8% more networks than the graph-based model. The results also show that, for the same RF environment, the hypergraph model requires up to 36% fewer channels than the graph model to achieve an average of 100% operational networks. The rate of growth of the running time of the hypergraph-based algorithm with respect to the input size is quadratic, like the graph-based algorithm.
There are two types of datasets included:
1) The network parameters of the 100 independent sets of 100 networks which were used in the two test scenarios:
a) Test senario 1 with networks of coverage radii 1 km, 5 km and 10 km in the ratio 40:35:25, respectively.
b) Test scenario 2 with networks of coverage radii 1 km and 10 km in the ratio 50:50, respectively.
2) Simulation results for Test 1 and Test 2. These results can be reproduced using the algorithms given in the paper and the data set of test networks as above.
a) results for number of operational networks for test scenario 1 when the number of available channels is 2.
b) results for number of operational networks for test scenario 1 when the number of available channels is 3.
c) results for the number of required channels for test scenario 1.
The results are from 100 independent simulations. The results are plotted into graphs using box and whisker plot to show the distribution of results.
Dataset Files
- Results for number of operational networks when K=2.xlsx (40.27 kB)
- Results for number of operational networks when K=3.csv (19.25 kB)
- Results for number of required channels.csv (4.19 kB)
- 100Networks_40_35_25.zip (147.32 kB)
- 100Networks_50_0_50.zip (145.60 kB)