Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Real-Time HAP-Assisted Vehicular Edge Computing for Rural Areas

Alessandro Traspadini, Marco Giordani, Giovanni Giambene, Michele Zorzi

TL;DR

The paper tackles real-time vehicular edge computing in rural areas by leveraging High Altitude Platforms as edge servers within Non-Terrestrial Networks. It introduces a queueing-theoretic model with Poisson frame arrivals and two processing paths (onboard via $M/D/1$ and HAP-assisted via $M/D/c$), and formulates an optimization over the offloading factor $\eta$ to maximize the real-time service probability $P_{\rm RT}$ under latency and capacity constraints. A closed-form-like treatment uses a geometric-tail approximation to obtain tractable waiting-time expressions, and the optimal $\eta^*$ is found with Brent optimization; results show substantial latency improvements from HAP offloading, with capacity thresholds (e.g., $C_{\rm HAP} \gtrsim 3000$ GFLOPS for $r \le 10$ fps) to guarantee real-time performance. The work provides practical guidance for designing rural VEC deployments with HAPs, balancing transmission delays and computational capacity to meet stringent latency requirements in autonomous sensing tasks.

Abstract

Non-Terrestrial Networks (NTNs) are expected to be a key component of 6th generation (6G) networks to support broadband seamless Internet connectivity and expand the coverage even in rural and remote areas. In this context, High Altitude Platforms (HAPs) can act as edge servers to process computational tasks offloaded by energy-constrained terrestrial devices such as Internet of Things (IoT) sensors and ground vehicles (GVs). In this paper, we analyze the opportunity to support Vehicular Edge Computing (VEC) via HAP in a rural scenario where GVs can decide whether to process data onboard or offload them to a HAP. We characterize the system as a set of queues in which computational tasks arrive according to a Poisson arrival process. Then, we assess the optimal VEC offloading factor to maximize the probability of real-time service, given latency and computational capacity constraints.

Real-Time HAP-Assisted Vehicular Edge Computing for Rural Areas

TL;DR

The paper tackles real-time vehicular edge computing in rural areas by leveraging High Altitude Platforms as edge servers within Non-Terrestrial Networks. It introduces a queueing-theoretic model with Poisson frame arrivals and two processing paths (onboard via and HAP-assisted via ), and formulates an optimization over the offloading factor to maximize the real-time service probability under latency and capacity constraints. A closed-form-like treatment uses a geometric-tail approximation to obtain tractable waiting-time expressions, and the optimal is found with Brent optimization; results show substantial latency improvements from HAP offloading, with capacity thresholds (e.g., GFLOPS for fps) to guarantee real-time performance. The work provides practical guidance for designing rural VEC deployments with HAPs, balancing transmission delays and computational capacity to meet stringent latency requirements in autonomous sensing tasks.

Abstract

Non-Terrestrial Networks (NTNs) are expected to be a key component of 6th generation (6G) networks to support broadband seamless Internet connectivity and expand the coverage even in rural and remote areas. In this context, High Altitude Platforms (HAPs) can act as edge servers to process computational tasks offloaded by energy-constrained terrestrial devices such as Internet of Things (IoT) sensors and ground vehicles (GVs). In this paper, we analyze the opportunity to support Vehicular Edge Computing (VEC) via HAP in a rural scenario where GVs can decide whether to process data onboard or offload them to a HAP. We characterize the system as a set of queues in which computational tasks arrive according to a Poisson arrival process. Then, we assess the optimal VEC offloading factor to maximize the probability of real-time service, given latency and computational capacity constraints.
Paper Structure (16 sections, 1 theorem, 20 equations, 2 figures)

This paper contains 16 sections, 1 theorem, 20 equations, 2 figures.

Table of Contents

  1. Introduction
  2. System Model
  3. Problem Formulation
  4. Delay Model
  5. Onboard processing
  6. HAP-assisted processing
  7. HAP-to-Ground Channel Model
  8. Queuing Model
  9. Optimal Offloading for HAP-Assisted VEC
  10. Performance evaluation
  11. Simulation Setup and Parameters
  12. Numerical Results
  13. Number of users (GVs)
  14. GV's computational capacity
  15. Frame rate
  16. ...and 1 more sections

Key Result

Lemma 1

Based on Eq. eq:p_j_approx, an empirical closed-form expression for the average waiting time of an M/D/$c$ queue with arrival rate $\lambda$ and service rate $\mu$ is given by where a relatively small value of $M$ can already provide an accurate approximation of $\bar{W}_q{({\text{M/D/}c})}$.

Figures (2)

  • Figure 1: Optimal offloading factor $\eta^*$ (right axis) and average latency $\bar{t}(\eta)$ (left axis) vs. $n$ and $C_{\rm GV}$, for $r=10$ fps and $n_\text{UL}=3$ Mb. Striped bars are interrupted to represent the case of unstable queues where the latency increases indefinitely in the long term.
  • Figure 2: Real-time probability (left) and average latency (right) vs. $C_{\rm HAP}$ and $r$, when $C_{\rm GV}=$ 800 GFLOPS, $n=100$ GVs and $n_\text{UL}=1$ Mb. Dash-dot lines and black striped bars refer to a baseline offloading scheme which balances the load of the HAP and of the GVs equally.

Theorems & Definitions (2)

  • Lemma 1
  • proof