Delay-Throughput Tradeoff for Supportive Two-Tier Networks

Long Gao, Rui Zhang, Changchuan Yin, Shuguang Cui

Abstract

This paper explores a cooperative paradigm in wireless networking where a dense secondary network (e.g., a cognitive radio network) actively assists a primary "legacy" network. By allowing the secondary network to relay packets for the primary tier, the primary network's throughput scaling can be significantly boosted from the classic $\Theta(1/\sqrt{n \log n})$ limit to $\Theta(1/\log n)$. Crucially, this "supportive" behavior is achieved without degrading the secondary network's own optimal delay-throughput performance.

Introduction

In large-scale wireless networks, the "Gupta-Kumar" limit represents a fundamental bottleneck: as the number of nodes $n$ increases, the throughput per node typically decays at a rate of $1/\sqrt{n}$. Traditional research into "two-tier" networks, where a primary licensed network and a secondary cognitive network share the same spectrum, has largely focused on interference avoidance. This paper shifts the focus toward a more ambitious goal: can the secondary network actually improve the primary network's performance?

The challenge lies in designing a protocol where the secondary tier acts as a high-density relay for the primary tier without requiring modifications to the primary network's legacy protocols and without sacrificing the secondary tier's own data transmission capabilities.

Methodology: The Supportive Multi-hop Protocol

The authors model a unit area with two tiers of nodes distributed according to Poisson Point Processes (PPP). The primary tier has density $n$, while the secondary tier has a much higher density $m = n^\beta$ (where $\beta \ge 2$). The core of the proposed approach is a Supportive Multi-hop Protocol characterized by spatial partitioning and time-division multiplexing.

Spatial and Temporal Management

The area is divided into primary and secondary cells. To manage interference, the protocol defines specific regions for transmission and reception.

Spatial partitioning and preservation regions in the supportive multi-hop protocol.

As shown in the figure above, the protocol utilizes "Preservation regions" around active primary nodes to ensure secondary transmissions do not cause harmful interference. Conversely, "Collection regions" are established to facilitate the gathering of primary packet segments by secondary nodes.

The secondary network's time frame is split into three subframes:

  1. Subframe 1: Secondary nodes transmit their own data.
  2. Subframe 2: Secondary nodes relay primary packet segments toward the destination in a parallel, multi-hop fashion.
  3. Subframe 3: Secondary nodes deliver the reassembled primary packets to the final primary destination.

Packet Splitting and Parallelism

When a primary node transmits, $N$ nearby secondary nodes capture the primary packet and divide it into $N$ smaller segments. By routing these segments through the dense secondary grid, the protocol bypasses the sparse distribution bottleneck of the primary tier. This process is transparent to the primary tier, which perceives the secondary nodes as highly efficient primary relays.

Key Results and Insights

The primary insight of this work is that the massive density of the secondary network ($\beta \ge 2$) can be leveraged as a resource to create parallel "highways" for primary data.

Throughput and Delay Scaling

The analysis reveals a significant performance gain for the primary tier:

  • Primary Throughput Boost: The throughput per S-D pair improves from $\Theta(1/\sqrt{n \log n})$ to $\Theta(1/\log n)$.
  • Primary Delay-Throughput Tradeoff: The new tradeoff for the primary tier is established as $D_p(n) = \Theta(\sqrt{n^\beta \log n} \lambda_p(n))$.
  • Secondary Stability: The secondary tier maintains its optimal stand-alone throughput scaling of $\Theta(1/\sqrt{m \log m})$ and its optimal delay-throughput tradeoff. The time-sharing overhead only introduces a constant factor penalty that does not affect asymptotic scaling.

Limitations and Future Directions

While the theoretical results are promising, several practical considerations remain for future work:

  • Mobility: The current model assumes static nodes. Node mobility typically improves throughput but increases delay, which would significantly alter the tradeoffs.
  • Density Constraints: The gains rely on the secondary network being significantly denser than the primary ($\beta \ge 2$).
  • Incentives: The model assumes the secondary tier is inherently cooperative. Future research could explore game-theoretic frameworks to provide incentives for cognitive networks to act supportively.
  • Channel Realism: Incorporating fading and shadowing into the pathloss model would provide a more realistic assessment of performance in complex environments.

Summary updated

BibTeX
@inproceedings{gao2009delay,
  author = {Long Gao and Rui Zhang and Changchuan Yin and Shuguang Cui},
  title = {{Delay-Throughput Tradeoff for Supportive Two-Tier Networks}},
  booktitle = {ISIT},
  year = {2009},
  doi = {10.1109/isit.2009.5205252},
}
Figure