[Paper Review] Novel Device-to-Device Discovery Scheme based on Random Backoff in LTE-Advanced Networks

Jiayi Zhang, Likai Deng+4 more|arXiv (Cornell University)|Jul 12, 2017

Advanced MIMO Systems Optimization9 references3 citations

TL;DR

This paper proposes a novel device-to-device (D2D) discovery scheme in LTE-Advanced networks using a random backoff mechanism to reduce collisions and improve discovery efficiency. By enabling D2D users to randomly select resource blocks and retransmit with backoff after collisions, the scheme significantly increases discovery probability and reduces average discovery delay compared to existing methods, with analytical and simulation validation.

ABSTRACT

Device-to-Device (D2D) discovery is a key enabler of D2D communications for the direct exchange of local area traffic between proximity users (UEs) to improve spectral efficiency. The direct D2D discovery relies on the capabilities of the D2D UEs to autonomously indicate their presence to proximity D2D UEs. {Despite} its potential of reducing energy and signalling burden, the direct D2D discovery {has not drawn adequate attention}. In this paper, we propose a direct D2D discovery scheme based on the random backoff procedure, where D2D UEs randomly choose a backoff interval and retransmit a beacon. Compared with existing schemes, the performance of the proposed scheme can be significantly enhanced in terms of the discovery probability and the discovery delay. Several useful guidelines for its design are proposed based on our analysis. Finally, numerical results provide valuable insights on the performance tradeoff inherent to the proposed D2D discovery scheme.

Motivation & Objective

• Address the lack of research on direct D2D discovery schemes that minimize signaling and energy overhead in LTE-Advanced networks.
• Overcome the limitations of network-assisted D2D discovery, such as high signaling load and battery consumption.
• Reduce collision probability and discovery delay in dense D2D environments where multiple users attempt to discover each other simultaneously.
• Design a scheme compatible with existing LTE-A infrastructure that leverages the random backoff mechanism used in random access procedures.
• Provide analytical and simulation-based guidelines for optimizing key parameters like backoff window size and maximum retransmissions.

Proposed method

• Adopt a distributed D2D discovery mechanism where each device independently selects a resource block (RB) to transmit a beacon.
• Implement a random backoff procedure: devices that experience a collision in the first transmission reselect a random backoff window and retransmit in a subsequent RB.
• Use the Balls & Bins combinatorics model to analytically evaluate collision probability, discovery success rate, and average discovery delay.
• Model the system using discrete time slots (DZs), where each D2D UE attempts discovery in a random RB within a given time window.
• Introduce two key control parameters: the maximum number of retransmissions $ L_{ ext{max}} $ and the backoff window size $ W $, to tune performance trade-offs.
• Derive closed-form expressions for the average number of discovered UEs and average discovery delay based on probabilistic analysis.

Experimental results

Research questions

• RQ1 How does the proposed random backoff-based D2D discovery scheme improve discovery probability compared to non-backoff schemes?
• RQ2 What is the impact of the backoff window size $ W $ on collision probability and discovery efficiency?
• RQ3 How does the maximum number of retransmissions $ L_{ ext{max}} $ affect the average discovery delay and number of discovered UEs?
• RQ4 What is the optimal balance between discovery delay and discovery success rate under varying network loads?
• RQ5 Can the proposed scheme achieve a target discovery probability by tuning $ W $ and $ L_{ ext{max}} $, and what are the performance trade-offs?

Key findings

• The proposed scheme increases the average number of discovered D2D UEs by up to 467 in a 200-DZ period with 1800 UEs, significantly outperforming existing schemes.
• The average discovery delay $ ar{D}_S $ is minimized at $ ar{D}_S = 0.186 $ when $ L_{ ext{max}} = 2 $, indicating optimal retransmission trade-off.
• Increasing the backoff window size $ W $ reduces collision probability, with the most pronounced benefit observed when the number of D2D UEs increases from 8 to 15.
• When $ L_{ ext{max}} $ increases from 2 to 7, the number of discovered UEs decreases due to higher retransmission overhead, but the gap between curves diminishes with higher $ L_{ ext{max}} $.
• For $ M ightarrow 2250 $ UEs, the average delay saturates at a constant value, while the number of discovered UEs continues to decline, indicating a performance ceiling under high load.
• The metric $ ar{D}_S / M_i(S) $ reaches a minimum of 0.186 at $ L_{ ext{max}} = 2 $, indicating this configuration optimizes the delay-to-discovery ratio.