16、UAV NOMA - MEC in IoT: Security Offloading and Optimization

UAV NOMA - MEC in IoT: Security Offloading and Optimization

1. Introduction

In the context of the Internet of Things (IoT), Unmanned Aerial Vehicle (UAV) - aided Non - Orthogonal Multiple Access (NOMA) with Mobile Edge Computing (MEC) has emerged as a promising approach for enhancing security offloading performance. This blog delves into the numerical results, system parameters, and the analysis of the Secrecy Success Cumulative Probability (SSCP) in such a system.

2. System Parameters

The following table presents the system parameters considered in all simulations:
| Parameter | Value | Parameter | Value | Parameter | Value |
| — | — | — | — | — | — |
| ((x_A, y_A)) | ((40, 0)) (m) | (A) | (0.1581) | (P) | ((0, 20)) (dB) |
| ((x_B, y_B)) | ((0, 40)) (m) | (B) | (9.6177) | (T) | (0.5) (s) |
| ((x_E, y_E)) | ((75, 75)) (m) | (O_{LoS}) | (1) | (\rho) | (0.75) |
| (x_U) | ((0, 50)) (m) | (O_{NLoS}) | (20) | (\sigma) | (2) |
| (y_U) | ((0, 50)) (m) | (c) | (3\times10^{8}) | (W) | (10^{8}) |
| (h_U) | ((20, 100)) (m) | (f_c) | (10^{7}) | (\beta_A=\beta_B) | (0.5) |
| (L) | (10^{4}) | (f_{MEC}^U) | (10^{8}) | (\mathcal{I}) | (10^{2}) |
| (O = Q) | (10^{2}) | (\varsigma) | (10^{2}) | (\mathcal{N}) | (100) |

3. Impact of Average SNR and Number of ED Clusters on SSCP

The impact of the average SNR (\gamma_0) and the number of Energy - harvesting Devices (ED) clusters on the SSCP of the entire system is shown in Figure 3.
- Monte Carlo Simulation and Analytical Match : The Monte Carlo simulation and the analytical results show a strong match, validating the accuracy of the proposed model.
- Effect of Device Number : When the number of devices in the two clusters increases, the SSCP also increases. This is because the UAV has more choices to select the best ED in the two clusters.
- Effect of Transmit Power : Increasing the transmit power of the device improves the system’s security offloading performance. As the transmit power rises, the device has more power to communicate with the UAV.

The following mermaid flowchart depicts the relationship between device number, transmit power, and SSCP:

graph LR
    A[Number of Devices in Clusters] -->|Increase| B[More UAV Selection Options]
    B -->|Leads to| C[Increase in SSCP]
    D[Transmit Power of Device] -->|Increase| E[More Power for Communication]
    E -->|Results in| C

4. Impact of UAV Altitude and Number of ED Clusters on SSCP

Figure 4 shows the impact of the UAV altitude (h_U) and the number of ED clusters on the SSCP of the entire system.
- Optimal Altitude : There exists an optimal altitude for the UAV to maximize the SSCP. When the UAV altitude is low, the probability of encountering Non - Line - of - Sight (NLoS) is greater than that of encountering Line - of - Sight (LoS) due to urban obstacles. As the altitude increases, the probability of LoS between the UAV and the ED becomes greater, improving the performance. However, a very high altitude increases the communication distance between the UAV and the ED, leading to increased path loss and decreased performance.

The following list summarizes the factors affecting SSCP based on UAV altitude:
1. Low altitude: Higher NLoS probability, lower SSCP.
2. Increasing altitude: Higher LoS probability, increasing SSCP.
3. Very high altitude: Increased path loss, decreasing SSCP.

5. Impact of UAV Location on SSCP

The 3D result in Figure 5 depicts the SSCP value domain as a result of the simultaneous effects of the UAV’s location ((x_U, y_U)).
- Optimal Location : There exists an optimal position ((x_U^ , y_U^ )) that optimizes the system’s performance. The UAV will select this optimal location to communicate with the EDs of the two clusters. This is an outstanding characteristic of UAV - based systems.

6. Impact of UAV Location and Altitude on SSCP

Figure 6 shows the impact of the UAV’s location and altitude ((x_U^ , y_U^ , h_U^ )) on the SSCP of the entire system.
- Optimization Algorithm *: By applying the SSCPMax - PSO algorithm to find the optimal values, the secret offloading performance is better compared to self - fixing the UAV’s parameters.

The following mermaid flowchart shows the process of optimizing SSCP using the SSCPMax - PSO algorithm:

graph LR
    A[Initial UAV Parameters] --> B[SSCPMax - PSO Algorithm]
    B --> C[Optimal UAV Location and Altitude \((x_U^*, y_U^*, h_U^*)\)]
    C --> D[Improved SSCP]

7. Conclusion

In summary, the study investigated the secrecy offloading performance of a UAV - aided NOMA - MEC system in IoT over a Rayleigh fading channel. A fourth - phase system operating protocol based on UAV - ED selection was proposed, focusing on NOMA - MEC techniques to enhance the secrecy offloading performance. Closed - form expressions of the SSCP of the entire system were obtained, and a PSO - based algorithm was proposed to determine the UAV’s location and altitude for maximizing the SSCP. Numerical results validated the proposed system’s secrecy offloading performance.

8. Proof of Theorem 1

By substituting equations (3), (10), (11), (13), (14), (15), (16), (17), (18) into (19), the system’s (\mathcal{S}) can be rewritten as:
[
\begin{align }
\mathcal{S}&=\Pr\left{X > \Delta_2(Y), Y > \Delta_1, P < \Delta_3(Y), Z < \left(\frac{\gamma_{A^ U}X}{\gamma_{B^ U}Y + 1}+1 - \theta_{A^ }\right)\frac{\gamma_{B^ E}P + 1}{\theta_{A^ }\gamma_{A^ E}}\right}\
&=\int_{\Delta_1}^{\infty}\int_{\Delta_2(Y)}^{\infty}\int_{0}^{\Delta_3(Y)}F_Z\left[\left(\frac{\gamma_{A^ U}X}{\gamma_{B^ U}Y + 1}+1 - \theta_{A^ }\right)\frac{\gamma_{B^ E}P + 1}{\theta_{A^ }\gamma_{A^ E}}\right]f_P(P)f_X(X)f_Y(Y)dPdXdY
\end{align }
]
where (X = |g_{A^ U}|^2), (Y = |g_{B^ U}|^2), (Z = |g_{A^ E}|^2), (P = |g_{B^ E}|^2), (\Delta_1=\frac{\varphi_{B^ }}{\gamma_{B^ U}}), (\Delta_2(Y)=\frac{\varphi_{A^ }(\gamma_{B^ U}Y + 1)}{\gamma_{A^ U}}), (\Delta_3(Y)=\frac{\gamma_{B^ U}Y + 1-\theta_{B^ }}{\theta_{B^ }\gamma_{B^*E}}).

8.1 Solving the First Integral ((I_1))

Combining the Cumulative Distribution Function (CDF) in (4) and the Probability Density Function (PDF) in (5) into (I_1), we have:
[
\begin{align }
I_1&=\frac{1}{\lambda_{B^ E}}\left(\int_{0}^{\Delta_3(Y)}e^{-\frac{P}{\lambda_Q}}dP - e^{-\Delta_5(X,Y)}\int_{0}^{\Delta_3(Y)}e^{-\Delta_4(X,Y)P}dP\right)\
&=1 - e^{-\frac{\Delta_3(Y)}{\lambda_{B^ E}}}-\frac{e^{-\Delta_5(X,Y)}}{\lambda_{B^ E}\Delta_4(X,Y)}\left(1 - e^{-\Delta_3(Y)\Delta_4(X,Y)}\right)
\end{align }
]
where (\Delta_4(X,Y)=\left(\frac{\gamma_{A^ U}X}{\gamma_{B^ U}Y + 1}+1 - \theta_{A^ }\right)\frac{\gamma_{B^ E}}{\lambda_{A^ E}\theta_{A^ }\gamma_{A^ E}}+\frac{1}{\lambda_{B^ E}}) and (\Delta_5(X,Y)=\left(\frac{\gamma_{A^ U}X}{\gamma_{B^ U}Y + 1}+1 - \theta_{A^ }\right)\frac{1}{\lambda_{A^ E}\theta_{A^ }\gamma_{A^*E}}).

8.2 Solving the Second Integral ((I_2))

Substituting (I_1) in (32) and the PDF in (8) into the second integral (I_2), we get:
[
\begin{align }
I_2&=\sum_{k = 1}^{N}\binom{N}{k}\frac{(- 1)^{k + 1}k}{\lambda_{A^ U}}\left[\int_{\Delta_2(Y)}^{\infty}e^{-\frac{kX}{\lambda_{A^ U}}}dX-\int_{\Delta_2(Y)}^{\infty}e^{-\frac{\Delta_3(Y)}{\lambda_{B^ E}}-\frac{kX}{\lambda_{A^ U}}}dX-\int_{\Delta_2(Y)}^{\infty}\frac{e^{-\Delta_5(X,Y)-\frac{kX}{\lambda_{A^ U}}}}{\lambda_{B^ E}\Delta_4(X,Y)}\left(1 - e^{-\Delta_3(Y)\Delta_4(X,Y)}\right)dX\right]
\end{align }
]
For (I_{21}), using Eq. (3.351.111), we have:
[
I_{21}=\frac{\lambda_{A^ U}}{k}e^{-\frac{k\Delta_2(Y)}{\lambda_{A^ U}}}-\frac{\lambda_{A^ U}}{k}e^{-\frac{\Delta_3(Y)}{\lambda_{B^ E}}-\frac{k\Delta_2(Y)}{\lambda_{A^ U}}}
]
For (I_{22}), by letting (v = e^{-X}) and (X=-\ln(v)) and applying the Gaussian - Chebyshev quadrature method:
[
I_{22}=\frac{\pi e^{-\Delta_2(Y)}}{2O}\sum_{o = 1}^{O}\sqrt{1-\zeta_o^2}\frac{\omega_o^{\frac{k}{\lambda_X}-1}e^{-\Delta_5(\delta_o,Y)}}{\lambda_{B^ E}\Delta_4(\delta_o,Y)}\left(1 - e^{-\Delta_3(Y)\Delta_4(\delta_o,Y)}\right)
]
where (\zeta_o=\cos\left(\frac{\pi(2o - 1)}{2O}\right)), (\omega_o=\frac{(\zeta_o + 1)e^{-\Delta_2(Y)}}{2}), (\delta_o=-\ln(\omega_o)), and (O) is the complexity versus accuracy trade - off coefficient.

Combining the results of (I_{21}) and (I_{22}), (I_2) can be rewritten as:
[
\begin{align }
I_2&=\sum_{k = 1}^{N}\binom{N}{k}(-1)^{k + 1}\left[e^{-\frac{k\Delta_2(Y)}{\lambda_{A^ U}}}-e^{-\frac{\Delta_3(Y)}{\lambda_{B^ E}}-\frac{k\Delta_2(Y)}{\lambda_{A^ U}}}-\frac{\pi k e^{-\Delta_2(Y)}}{2O\lambda_{A^ U}\lambda_{B^ E}}\sum_{o = 1}^{O}\sqrt{1-\zeta_o^2}\frac{\omega_o^{\frac{k}{\lambda_{A^ U}}-1}e^{-\Delta_5(\delta_o,Y)}}{\Delta_4(\delta_o,Y)}\left(1 - e^{-\Delta_3(Y)\Delta_4(\delta_o,Y)}\right)\right]
\end{align }
]

8.3 Solving the Final Integral

Finally, combining (I_2) in (36) and the PDF in (8) into the last integral in (30), (\mathcal{S}) is expressed as:
[
\begin{align }
\mathcal{S}&=\sum_{u = 1}^{M}\binom{M}{u}\frac{(-1)^{u + 1}u}{\lambda_{B^ U}}\sum_{k = 1}^{N}\binom{N}{k}(-1)^{k + 1}\left[e^{-\Xi_2(k)}\int_{\Delta_1}^{\infty}e^{-\Xi_1(k,u)Y}dY - e^{-\Xi_4(k)}\int_{\Delta_1}^{\infty}e^{-\Xi_3(k,u)Y}dY-\frac{\pi k}{2O\lambda_{A^ U}\lambda_{B^ E}}\sum_{o = 1}^{O}\sqrt{1-\zeta_o^2}\int_{\Delta_1}^{\infty}e^{-\Delta_2(Y)}\frac{\omega_o^{\frac{k}{\lambda_{A^ U}}-1}e^{-\Delta_5(\delta_o,Y)-\frac{uY}{\lambda_{B^ U}}}}{\Delta_4(\delta_o,Y)}\left(1 - e^{-\Delta_3(Y)\Delta_4(\delta_o,Y)}\right)dY\right]
\end{align*}
]

9. Practical Implications and Considerations

9.1 Deployment Strategies

When deploying a UAV - aided NOMA - MEC system in an IoT environment, the following practical strategies can be considered based on the above findings:
- Device Clustering : Grouping energy - harvesting devices into clusters can significantly improve the SSCP. Operators should aim to increase the number of devices in each cluster to provide the UAV with more options for selecting the best - performing EDs.
- UAV Altitude and Location : The optimal altitude and location of the UAV are crucial for maximizing the SSCP. Before deployment, a detailed analysis of the environment, including the presence of urban obstacles, should be conducted. The SSCPMax - PSO algorithm can be used to determine the best altitude and location for the UAV.

The following table summarizes the deployment strategies and their impacts:
| Deployment Strategy | Impact on SSCP |
| — | — |
| Increase device clustering | Increases SSCP due to more UAV selection options |
| Optimize UAV altitude | Maximizes SSCP by balancing LoS and path loss |
| Optimize UAV location | Improves SSCP by ensuring better communication with EDs |

9.2 System Scalability

As the number of IoT devices continues to grow, the scalability of the UAV - aided NOMA - MEC system becomes an important consideration.
- Increasing Device Density : With more devices in the system, the UAV can select from a larger pool of EDs, potentially increasing the SSCP. However, this also requires more computational resources for the UAV to perform the selection process.
- Multiple UAVs : In large - scale IoT deployments, multiple UAVs can be used to cover a wider area. Each UAV can be responsible for a specific set of ED clusters, and the overall SSCP can be optimized by coordinating the altitudes and locations of all UAVs.

The following mermaid flowchart shows the process of system scalability:

graph LR
    A[Increasing IoT Devices] -->|More Selection Options| B[Potential Increase in SSCP]
    A -->|Higher Computational Demand| C[Resource Management]
    D[Large - scale IoT Deployment] -->|Multiple UAVs| E[Coordinated UAV Operation]
    E -->|Optimize Overall SSCP| B

10. Future Research Directions

10.1 Advanced Optimization Algorithms

Although the PSO - based algorithm shows promising results in optimizing the UAV’s location and altitude, there is room for improvement. Future research could explore more advanced optimization algorithms, such as genetic algorithms or ant colony optimization, to further enhance the SSCP.

10.2 Integration with Other Technologies

The UAV - aided NOMA - MEC system can be integrated with other emerging technologies in the IoT ecosystem.
- Artificial Intelligence : AI techniques can be used to predict the behavior of IoT devices and optimize the UAV’s operation in real - time. For example, machine learning algorithms can analyze historical data to predict the best time to offload tasks from EDs to the UAV.
- Blockchain : Blockchain technology can enhance the security and trustworthiness of the offloading process. By using blockchain, the transactions between the UAV, EDs, and the MEC server can be recorded and verified, ensuring the integrity of the data.

The following list outlines the future research directions:
1. Develop advanced optimization algorithms for UAV operation.
2. Integrate the system with AI for real - time optimization.
3. Incorporate blockchain technology for enhanced security.

11. Conclusion

In conclusion, the UAV - aided NOMA - MEC system in IoT offers a promising solution for improving security offloading performance. Through numerical analysis and theoretical proof, we have demonstrated the importance of factors such as device clustering, UAV altitude, and location in maximizing the SSCP. The proposed PSO - based algorithm provides an effective way to optimize the UAV’s operation. However, there are still many areas for future research, including advanced optimization algorithms and integration with other technologies. By continuing to explore these directions, we can further enhance the performance and security of IoT systems.

To summarize the key points of this blog:
1. The SSCP of the system is affected by factors such as device number, transmit power, UAV altitude, and location.
2. The proposed PSO - based algorithm can optimize the UAV’s location and altitude to maximize the SSCP.
3. Practical deployment strategies and future research directions have been discussed to improve the system’s performance and scalability.

graph LR
    A[Key Factors] --> B[SSCP Optimization]
    B --> C[Practical Deployment]
    C --> D[Future Research]
    D -->|Enhance Performance| B

This flowchart shows the overall relationship between key factors, optimization, deployment, and future research in the UAV - aided NOMA - MEC system for IoT.