Optimal Context Aware Transmission Strategy for non-Orthogonal D2D Communications

TL;DR

This paper develops a rigorous theoretical analysis to quantify the balance between the gain offered by a D2D transmission, and its impact on the other network communications, and derives two theorems that define the optimal strategy to be employed, in terms of throughput maximization, when a single or multiple transmit power levels are available for the D1D communications.

Abstract

The increasing traffic demand in cellular networks has recently led to the investigation of new strategies to save precious resources like spectrum and energy. A possible solution employs direct device-to-device (D2D) communications, which is particularly promising when the two terminals involved in the communications are located in close proximity. The D2D communications should coexist with other transmissions, so they must be careful scheduled in order to avoid harmful interference impacts. In this paper, we analyze how a distributed context-awareness, obtained by observing few local channel and topology parameters, can be used to adaptively exploit D2D communications. We develop a rigorous theoretical analysis to quantify the balance between the gain offered by a D2D transmission, and its impact on the other network communications. Based on this analysis, we derive two theorems that define the optimal strategy to be employed, in terms of throughput maximization, when a single or multiple transmit power levels are available for the D2D communications. We compare this strategy to the state-of-the-art in the same network scenario, showing how context awareness can be exploited to achieve a higher sum throughput and an improved fairness.

Figures (9)

• Figure 1: Evolution of the MDP when $\{p\}$ and $\{q\}$ are stationary and non time correlated. The states on the left are those with $\Lambda_t = 0$, each one identified by the value of $P_t$ and $Q_t$. Dashed arrows correspond to action $T$, while solid arrows to action $H$. Here it is assumed that $|\mathcal{P}| = |\mathcal{Q}| = 4$ and $W=3$.
• Figure 2: Scenario: D2D communication in a cellular network, with one licensed user $U$ transmitting to the BS $B$, and one additional user $S$ attempting a D2D transmission to $D$.
• Figure 3: The optimal policy for the scenario with the following coordinates: $B = (0,0)$, $S = (100, 0)$, $D = (100, 80)$ and $U = (0, 120)$. There are 4 power levels $P_i = 0.05\times2^{i-1}\,\,\,\!\! W$, with $i\in\{1,2,3,4\}$. In case (a) we have $W = 10$, whereas in (b) $W = 3$, resulting in $k = 0.5571$ and $k = 1.1613$, respectively.
• Figure 4: The average throughput of $U$, as a function of the target SNR $\xi$ that is set by $S$.
• Figure 5: Probability of $S$ choosing the D2D transmission mode, as a function of both $W$ and $\xi$.

Theorems & Definitions (15)

• Lemma III.1
• Theorem III.1
• proof
• Lemma V.1
• Lemma V.2
• Lemma V.3
• Remark V.1
• Remark V.2
• Remark V.3
• Theorem V.1