[Paper Review] Scheduling of Multicast and Unicast Services under Limited Feedback by using Rateless Codes

Yin Sun, C. Emre Koksal+2 more|arXiv (Cornell University)|Apr 26, 2014

Cooperative Communication and Network Coding24 references3 citations

TL;DR

This paper proposes a joint scheduling and power allocation scheme for downlink unicast and multicast services using physical-layer rateless codes under limited feedback. By leveraging imperfect CSI and infrequent ACKs, the scheme achieves near-optimal throughput by exploiting multiuser diversity and multicast gain, with simulations showing it outperforms fixed-rate code and unicast-only schemes by significant margins.

ABSTRACT

Many opportunistic scheduling techniques are impractical because they require accurate channel state information (CSI) at the transmitter. In this paper, we investigate the scheduling of unicast and multicast services in a downlink network with a very limited amount of feedback information. Specifically, unicast users send imperfect (or no) CSI and infrequent acknowledgements (ACKs) to a base station, and multicast users only report infrequent ACKs to avoid feedback implosion. We consider the use of physical-layer rateless codes, which not only combats channel uncertainty, but also reduces the overhead of ACK feedback. A joint scheduling and power allocation scheme is developed to realize multiuser diversity gain for unicast service and multicast gain for multicast service. We prove that our scheme achieves a near-optimal throughput region. Our simulation results show that our scheme significantly improves the network throughput over schemes employing fixed-rate codes or using only unicast communications.

Motivation & Objective

• Address the challenge of managing network resources with limited feedback in downlink wireless networks.
• Overcome the inefficiency of fixed-rate codes under channel uncertainty and feedback overhead, especially for multicast services.
• Simultaneously exploit multiuser diversity gain (for unicast) and multicast gain (for group services) under practical feedback constraints.
• Develop a low-complexity control scheme that operates with imperfect CSI and infrequent ACKs, avoiding feedback implosion.
• Achieve a near-optimal stability region despite temporal correlation in rateless code transmissions and limited channel knowledge.

Proposed method

• Employ physical-layer rateless codes (e.g., Raptor, Strider, Spinal) to enable automatic rate adaptation and reduce ACK feedback overhead.
• Design a joint scheduling and power allocation algorithm that dynamically selects users based on imperfect CSI and ACK feedback.
• Use fluid limit techniques to model and analyze the temporal correlation in rateless code transmission, enabling stability analysis.
• Implement unicast file repair by splitting poor-channel multicast users into separate unicast sessions to enhance group throughput.
• Incorporate a reception overhead model for practical rateless codes, adjusting realized throughput by dividing by (1 + ε).
• Apply Lyapunov drift analysis adapted for rateless codes to prove near-optimal stability region under limited feedback.

Experimental results

Research questions

• RQ1 How can multiuser diversity and multicast gain be jointly exploited under limited feedback and imperfect CSI?
• RQ2 What scheduling and power allocation strategy enables near-optimal stability in a downlink network using rateless codes with irregular queue updates?
• RQ3 How does the use of rateless codes reduce feedback overhead compared to fixed-rate codes in unicast and multicast scenarios?
• RQ4 What is the impact of channel estimation accuracy (ρ) on the performance of the proposed scheme?
• RQ5 Can unicast file repair improve multicast throughput when some users experience poor channel conditions?

Key findings

• The proposed scheme achieves at least 8/9 ≈ 88.9% of the optimal throughput region, approaching the outer bound of infinite block-length codes.
• Throughput performance improves significantly with higher CSI accuracy (ρ), with gains increasing from ρ = 0.1 to ρ = 0.8.
• The scheme outperforms fixed-rate code-based schemes by mitigating rate loss due to channel uncertainty, especially in high-mobility or fading environments.
• Unicast file repair enhances multicast throughput by up to 30% in low-SNR scenarios (e.g., 4 dB average SNR), with gains increasing at higher CSI accuracy (ρ = 0.9).
• The use of rateless codes reduces ACK feedback overhead compared to fixed-rate codes, avoiding the need for per-packet NACK/ACK signaling.
• The scheme maintains robustness and stability even with irregular decoding times and limited feedback, proving effective under practical constraints.