System Throughput

---
name: system-throughput
description: This evaluation probes the trade-off between total system throughput and user fairness in UAV-enabled wireless networks. It measures how effectively a resource allocation and trajectory design scheme balances maximizing aggregate data rates against ensuring equitable service across users with varying channel conditions. Use when the user has predictions and gold and needs to compute system throughput.
metadata:
  skill_kind: metric
  source_arxiv: 2406.04750
  bibtex_key: [REDACTED]
  confidence: high
---

# system-throughput

> Throughput and Fairness Trade-off Balancing for UAV-Enabled Wireless Communication Systems — Kejie Ni et al. (2024) (arXiv:2406.04750, 2024)

## What this evaluates

This evaluation probes the trade-off between total system throughput and user fairness in UAV-enabled wireless networks. It measures how effectively a resource allocation and trajectory design scheme balances maximizing aggregate data rates against ensuring equitable service across users with varying channel conditions.

## Datasets

- **UAV Wireless Communication Simulation** — total ?; splits: test (-1)

## Metrics

- `system throughput` **(primary)** — range: bps
  - Average throughput of K users over N time slots, calculated as the sum of per-user per-slot throughputs divided by the total number of user-time slots.
- `variance of throughput` — range: other
  - Statistical variance of the average throughput across all users, used to quantify the stability and fairness of resource allocation over time.

## Input / output format

**Input**: User locations (K=9 on horizontal plane), UAV altitude (500m), max speed (40m/s), bandwidth (10MHz), noise power (-169 dBm/Hz), max transmit power (0.1W), channel gain at 1m (-50dB), Rician fading parameters, flight period (50s), time slots (50), and fairness factor α.

**Output**: UAV trajectory, bandwidth and power allocation per user per time slot, and resulting per-user throughput R_k[n] for each time slot.

## Scoring recipe

```python
def compute_metrics(R_k_n, K, N):
    # R_k_n is a matrix of shape (K, N) containing throughput per user per slot
    system_throughput = sum(R_k_n[k][n] for k in range(K) for n in range(N)) / (K * N)
    user_avg = [sum(R_k_n[k][n] for n in range(N)) / N for k in range(K)]
    mean_user = sum(user_avg) / K
    variance_throughput = sum((u - mean_user)**2 for u in user_avg) / K
    return system_throughput, variance_throughput
```

## Common pitfalls

- The fairness factor α must be tuned to ensure convexity of the optimization constraints (α R_k[n] ≤ 1), which limits its valid range and affects convergence.
- Benchmark schemes are evaluated under the most stringent QoS constraints, which may overstate their performance compared to unconstrained baselines.
- Throughput is highly sensitive to user density; fewer users yield higher throughput but reduce the dynamic range of α that guarantees convex constraints.

## Evidence (verbatim from paper)

> In Fig. 4 (a), we analyze system throughput (i.e. average throughput of K users over N time slots) of the developed UAV communication system. ... We use the variance of throughput as a metric to quantify the stability of throughput as depicted in Fig. 5 (b).

## Citation

```bibtex
@misc{ni2024throughput,
  title={Throughput and Fairness Trade-off Balancing for UAV-Enabled Wireless Communication Systems},
  author={Kejie Ni et al. (2024)},
  year={2024},
  note={arXiv:2406.04750}
}
```

- arXiv: 2406.04750