Throughput and Delay Scaling in Supportive Two-Tier Networks

Long Gao, Rui Zhang, Changchuan Yin, Shuguang Cui

TL;DR

This paper introduces the concept of a "supportive" two-tier wireless network where a secondary (cognitive) tier can actively relay data for a primary (legacy) tier. The authors demonstrate that this supportive interaction can significantly improve the primary tier's throughput scaling, even achieving better performance than a standalone network, while the secondary tier maintains its own network performance.

What Problem Does This Paper Address?

Modern wireless networks are increasingly complex, with the emergence of cognitive radio technologies enabling secondary users to opportunistically access spectrum. While previous work has explored two-tier networks (primary and secondary users sharing spectrum), these often treated them as non-interactive. This paper addresses the question: can a secondary network actively assist a primary network to improve its performance, specifically its throughput and delay scaling, without compromising the secondary network's own capabilities? The authors investigate this by proposing a model where secondary users can act as relays for primary users, going beyond simple spectrum sharing. This is particularly relevant for scenarios where a legacy primary network needs performance enhancements without a complete overhaul, by introducing a new cognitive layer.

Methods

The core of this paper lies in proposing and analyzing novel protocols for a supportive two-tier network. The authors define two scenarios: static primary and secondary nodes, and static primary nodes with mobile secondary nodes.

  • Primary Protocol: A modified time-slotted multi-hop transmission scheme is used. Key elements include:
  • Dividing the network into primary cells and clusters.
  • A designated relay node within each primary cell (which can be a secondary node).
  • A handshake mechanism for packet transmission and acknowledgment.
  • A specific data path routing strategy.
  • Secondary Protocol: This protocol is designed to opportunistically assist the primary tier.
  • Static Case: Secondary nodes chop primary packets into smaller pieces, reassemble them, and deliver them to primary destinations. They act as designated relay nodes.
  • Mobile Case: Secondary nodes store primary packets and deliver them when they move into the vicinity of the primary destination. This involves defining "preservation regions" to avoid interfering with primary transmissions and "collection regions" for delivery.
  • Mobility Models: Two models for secondary nodes are considered: independent and identically distributed (i.i.d.) mobility and a random walk (RW) model.

The analysis focuses on throughput and delay scaling laws, using mathematical derivations based on these protocols and network densities.

Datasets

Not discussed in the paper.

Key Insights and Ideas

  • Supportive Relaying is Key: The most significant insight is that allowing secondary users to actively relay primary traffic, rather than just sharing spectrum, can lead to substantial performance gains for the primary network. This is a departure from non-supportive two-tier models.
  • Cognitive Radio as an Enabler: The cognitive nature of secondary users is crucial, allowing them to opportunistically access spectrum and act as relays without needing pre-allocated primary spectrum resources.
  • Throughput Improvement: The primary tier can achieve a per-node throughput scaling of $\Theta(1/\log n)$, a significant improvement over the $\Theta(1/\sqrt{n \log n})$ scaling in a standalone static network.
  • Mobility Benefits: Mobile secondary nodes offer even better performance for the primary tier, especially in terms of delay. The i.i.d. mobility model leads to a constant delay scaling ($\Theta(1)$), a remarkable achievement.
  • Secondary Tier Autonomy: Despite assisting the primary tier, the secondary tier can still achieve throughput and delay scaling laws comparable to those of a standalone secondary network.
  • Trade-offs Exist: While supportive relaying improves throughput, it can sometimes lead to worse delay performance for the primary tier in certain scenarios (e.g., static secondary nodes) compared to a standalone network, highlighting a throughput-delay trade-off.

Key Results

  • Static Scenario:
    • Primary Tier Throughput: $\lambda_p(n) = \Theta(1/\log n)$
    • Primary Tier Delay: $D_p(n) = \Theta(n \log n / \lambda_p(n))$
    • Secondary Tier Throughput: $\lambda_s(m) = \Theta(1/\sqrt{m \log m})$
    • Secondary Tier Delay: $D_s(m) = \Theta(m \lambda_s(m))$
  • Mobile Scenario (i.i.d. Mobility):
    • Primary Tier Throughput: $\lambda_p(n) = \Theta(1/\log n)$
    • Primary Tier Delay: $D_p(n) = \Theta(1)$
  • Mobile Scenario (RW Mobility):
    • Primary Tier Throughput: $\lambda_p(n) = \Theta(1/\log n)$
    • Primary Tier Delay: $D_p(n) = \Theta(1/S)$, where S is the RW step size.

The paper shows that the primary tier can achieve a better throughput scaling law than a standalone network, and in the mobile secondary node case, the delay scaling is also significantly improved.

Limitations and Future Work

  • Assumption $\beta \ge 2$: The analysis for the primary tier's throughput and delay scaling relies on the assumption that the density of secondary nodes ($\beta$) is at least twice that of primary nodes. The authors mention that the case $1 < \beta < 2$ is left for future work.
  • Constant Power: The analysis assumes constant transmit power for both tiers.
  • Simplified Channel Model: The paper considers a large-scale pathloss model and ignores shadowing and small-scale fading.
  • Security Issues: Potential security issues between the two tiers are explicitly not considered.
  • Future Work: The authors suggest extending the analysis to the $1 < \beta < 2$ regime and exploring more complex mobility models or channel conditions.

Summary updated

BibTeX
@article{gao2012throughput,
  author = {Long Gao and Rui Zhang and Changchuan Yin and Shuguang Cui},
  title = {{Throughput and Delay Scaling in Supportive Two-Tier Networks}},
  journal = {IEEE J. Sel. Areas Commun.},
  year = {2012},
  url = {https://openreview.net/forum?id=CfQlQPFSHnd},
}
Figure