Mosaba Simulation Eval

---
name: mosaba-simulation-eval
description: Evaluates a peer-to-peer wireless power transfer (P2P-WPT) framework's ability to balance energy across a mobile crowd while minimizing energy loss and maximizing network energy retention. It probes how well mobility and social-aware peer selection algorithms perform in dynamic, simulated environments. Use when the user wants to benchmark on MoSaBa Simulation Scenario, or asks about evaluating this task. Reports Total network energy.
metadata:
  skill_kind: dataset_eval
  source_arxiv: 2204.09399
  bibtex_key: ojha2022balanced
  confidence: high
---

# mosaba-simulation-eval

> Balanced Wireless Crowd Charging with Mobility Prediction and Social Awareness — Ojha et al. (2022) (arXiv:2204.09399, 2022)

## What this evaluates

Evaluates a peer-to-peer wireless power transfer (P2P-WPT) framework's ability to balance energy across a mobile crowd while minimizing energy loss and maximizing network energy retention. It probes how well mobility and social-aware peer selection algorithms perform in dynamic, simulated environments.

## Datasets

- **MoSaBa Simulation Scenario** — total ?; splits: test (-1)

## Metrics

- `Total network energy` **(primary)** — range: other
  - Sum of remaining energy of all nodes in the network at each iteration. Higher values indicate less energy loss during P2P-WPT exchanges.
- `Total energy variation distance` — range: other
  - Measures the variation distance in energy levels between nodes at each iteration. Lower values indicate higher energy balance quality.
- `Number of P2P meetings` — range: other
  - Count of peer pairs engaging in energy exchange per iteration.
- `Number of nodes that reach energy balance` — range: other
  - Count of nodes achieving the expected energy target level.
- `Execution time` — range: other
  - Actual CPU execution time per iteration for the method.

## Input / output format

**Input**: Simulated environment state including user locations, mobility patterns, social connections, initial energy levels [0,100], and current iteration step.

**Output**: Peer selection decisions (which users pair for P2P-WPT) and resulting energy states per iteration.

## Scoring recipe

```python
def evaluate_iteration(users, peer_pairs, alpha=0.5, beta=0.3):
    total_energy = sum(u.energy for u in users)
    variation = std([u.energy for u in users])
    meetings = len(peer_pairs)
    for u1, u2 in peer_pairs:
        exchanged = min(u1.energy, u2.energy) * alpha
        loss = exchanged * beta
        u1.energy -= loss
        u2.energy -= loss
    return total_energy, variation, meetings
```

## Common pitfalls

- The evaluation relies on artificially generated mobility and social data rather than real-world traces, limiting generalizability to actual crowd dynamics.
- Energy loss rate (beta) and charging rate (alpha) are fixed or varied within narrow ranges in experiments, which may not reflect diverse real-world hardware capabilities.
- Metrics are averaged over 50 runs for statistical smoothness, but variance or confidence intervals are not explicitly reported in the figures.

## Evidence (verbatim from paper)

> Total network energy refers to the total network energy of all the nodes in the experiment at each iteration. This metric computes the remaining energy of the network in each iteration and helps us understand the energy loss occurring in each method. It is important to note that energy loss also increases with increased amount of energy exchanged.

## Citation

```bibtex
@misc{ojha2022balanced,
  title={Balanced Wireless Crowd Charging with Mobility Prediction and Social Awareness},
  author={Ojha et al. (2022)},
  year={2022},
  note={arXiv:2204.09399}
}
```

- arXiv: 2204.09399