Normalized RB over Sample Clauses

Normalized RB over utilization avoidance optimization compared to average utilization agnostic and worst case scenarios. 86

Normalized RB over utilization avoidance optimization compared to av- erage utilization agnostic and worst case scenarios. the worst case where the under study D2D links would be assigned with the over-utilized RBs. As depicted, for example in the case of hexagonal deployment, the proposed method succeeds in utilizing less used RBs and outperforms the worst case as well as the average utilization agnostic case in a percentage of 45% and almost 28% over-utilization avoidance gain, respectively. Specifically, this performance gain can be depicted in the same figure for the case of more than 60% of overall network congestion. We note that similar behaviour is observed for more congested instances (i.e. > 70% and > 80%) where the achieved gain remains significant. It has to be also noted that, because the number of distributed D2D links as well as the input values ρ (already assigned RBs) are the same for both hexagonal and PPP-Voronoi tessellation layouts, the differences in terms of RB over-utilization are negligible between each other. Evaluation on MOCA-III Problem (3.17) aims at efficiently performing resource-aware balancing while tak- ing into account the existence of potential cellular interferers to the D2D commu- nications. In that sense, the interference-aware part adds a useful decision-making dimension to the optimization setting and mainly relates to the statistical chance of a link to interfere with a CUE. In essence, it chooses the association of a D2D link with the BS that will result in the lowest probability of interfering with a closely located CUE. Figure3.4 represents the interference region of a CUE (orange-dashed circular area in the figure) as the area under which cellular users might interfere with a DUE receiver. In specific, this area can be specified by the transmitting CUE’s location (centre of the circle) and a radius that is de- fined as the range of the interference. For approximation, the interference range for each CUE might vary and can be approximated by the multiplication of the respective cellular link range with a random variable β ≥ 1 [70]. Without loss of generality, we consider an upper limit for this variable and assume that β ∈ [1, 2]. By considering uniform distribution for both CUEs and DUEs, it might be pos- sible to have an equal number of potential cellular interferers when a D2D link associates with two different BSs. A representative toy example is shown in Fig- ure 3.7: we consider a cell-edge D2D link that its transmitter is ...