Risk Evaluation of Embedded Linux in Aerospace based on Cloud Model

doi:10.23940/ijpe.18.12.p31.32203227

Risk Evaluation of Embedded Linux in Aerospace based on Cloud Model

Yu Su^,^a, Yushuai Liu^a^,^b, Li Sun^a^,^b, Zhexi Yao^a^,^b, and Jinbo Wang^a

a Technology and Engineering Center for Space Utilization, Chinese Academy of Sciences, Beijing, 100094, China

b University of Chinese Academy of Sciences, Beijing, 100094, China

^*Corresponding Author(s): * E-mail address: suyu@csu.ac.cn

Accepted: ; Published:

Abstract

For the fuzziness and randomness of risk assessment of aerospace embedded Linux, a comprehensive assessment method of aerospace embedded Linux based on cloud model theory is proposed. Cloud model was used to replace the traditional membership function. Seven factors of embedded Linux were selected as the risk assessment set, and the risk level was established. By using test data and expert scoring cameras, the weight coefficient matrix and comprehensive evaluation matrix are constructed by the reverse cloud generator, and the comprehensive evaluation grade was obtained by the forward cloud generator. Taking Linux based on PREEMPT-RT patch as an example, the risk assessment method of space embedded Linux based on cloud model was validated. The results show that the assessment method is effective and feasible.

Keywords: cloud model; risk assessment; comprehensive fuzzy evaluation method; membership function

PDF (459KB) Metadata Related articles Export EndNote| Ris| Bibtex

Cite this article

Yu Su, Yushuai Liu, Li Sun, Zhexi Yao, Jinbo Wang. Risk Evaluation of Embedded Linux in Aerospace based on Cloud Model. International Journal of Performability Engineering, 2018, 14(12): 3220-3227 doi:10.23940/ijpe.18.12.p31.32203227

1. Introduction

In recent years, countries around the world have gradually realized the enormous role of manned space technology in national production and national defense. Spacecraft for various purposes emerge endlessly, especially the commercial space group represented by SPACE-X, which has constantly updated various launch records. China had launched Tiangong-1 and Tiangong-2 space laboratories in 2011 and 2016. At present, the manned space station is under development. The size and complexity of the spacecraft and its internal loads have increased unprecedentedly. VxWorks has excellent performance and has successfully carried out manned space missions for many times. However, the price is high and it will also bring security problems due to no open source. Embedded real-time Linux is gradually used in aerospace instead of Vx works in many fields because of its open source, rich interface and good real-time performance. In January 2014, JAXA, the Japan Aerospace Research and Development Agency, transferred the satellite control platform of the Sun Observation Satellite “Sunrise” and the observation satellite GEOTAIL to the Linux platform [1]. In fact, there are many real-time schemes for embedded Linux, such as Xenomai, RTAI, Preempt-RT, RT-Linux and so on. How to use scientific methods to evaluate several real-time Linux implementations scientifically and systematically and choose the best real-time Linux as the spacecraft load control system for different application scenarios has become an urgent problem to be solved. The new requirements of reliability and compatibility have important practical significance for the evaluation of space embedded real-time Linux.

There are many evaluation methods, such as Analytic Hierarchy Process (AHP)[2-4], Artificial Neural Network (ANN)[5] and Cloud Model(CM)[6], etc. The membership function of AHP is restricted by the cultural background of the assessor. It is very difficult to test the consistency of the judgment matrix, and the criterion CR< 0.1 is lack of scientific basis[7]. At the same time, the biggest problem of AHP is that it is difficult to guarantee the consistency of thinking when the evaluation index of a certain level reaches four or more. Because of the complexity of operating system evaluation index and the slow convergence speed, artificial neural network method cannotbe applied to the comprehensive evaluation system of operating system.

The evaluation of aerospace embedded real-time Linux is a comprehensive multi-factor evaluation, with the characteristics of fuzziness and randomness. Therefore, taking manned space engineering as the background, a real-time Linux evaluation system for aerospace embedded system based on cloud model is proposed in this paper. Finally, the effectiveness is verified by specific experiments.

2. Cloud Model Theory

2.1. Cloud Model Definition and Digital Characteristics

Cloud model is a two-way cognitive model between a qualitative concept expressed by linguistic value and its quantitative representation. Let U be a quantitative universe represented by exact numerical values, and C be a qualitative concept on U. If the quantitative value x∈U, and x is a random realization of the qualitative concept C, the determinacy of x to C is a random number with a stable tendency, then the distribution of x on U is called a cloud, and each x is called a cloud droplet (x, μ(x))[8-9].

Clouds consist of cloud droplets. Each cloud droplet is a point of qualitative concept mapping to the number field space. Cloud models generally represent a qualitative concept by three numerical features: expected value (E_x), entropy(E_n) and hyper entropy(H_e)[10-11]. Expectation E_x is the most representative point in qualitative concepts; Entropy En is the quantity representing the uncertainty of qualitative concepts. It is not only the reflection of the discrete degree of cloud droplets in qualitative concepts, but also the measurement of the range of cloud droplet values acceptable to the concepts in the universe. Super Entropy He reflects the condensation degree of cloud droplets in cloud models. The digital characteristics of the cloud model are shown in Figure 1.

Figure 1

New window| Download| PPT slide

Figure 1. The digital characteristics of the cloud model

2.2. Cloud Generator

The cloud generator is divided into a cloud forward generator and a cloud reverse generator. Forward Cloud Generator [12] is a forward and direct process in which qualitative concepts are transformed into quantitative values. Expectations, entropy, superb entropy and cloud droplet number of cloud models are input into the generator to obtain the quantitative values of cloud droplets in the domain and the determinacy of their representation concepts. If the random number x conforms to the normal distribution with the mean E_x and the standard deviation E_n, then for the normal random number N, the normal distribution with the mean E_x and the standard deviation H_e is satisfied, and the determinacy of the qualitative concept C by X satisfies:

(1)

μ (x) =

exp[-(x-E_x)²/(2N²)]

Reverse Cloud Generator[13] is a reverse and indirect process that converts quantitative values into qualitative concepts. It calculates cloud model expectations, entropy andsuperb entropy by using statistical “cloud droplets” and provides an effective way for fuzzy comprehensive evaluation. First, by calculating the mean $\overset{̅}{x}$ and the variance s² of the sample, we get the expectation, entropy and excess entropy of the cloud model, as shown below.

(2)

\{\begin{matrix} E_{x} = \overset{̅}{x} \\ E_{n} = \frac{\sqrt[]{\frac{π}{2}}}{n} \sum_{1}^{n} | x - E | \\ H_{e} = \sqrt[]{s^{2} - E_{n}^{2}} \end{matrix}

3. Evaluation Steps

The general steps of the comprehensive evaluation method based on cloud model proposed in this paper are as follows:

1) Analyze the relationship between the evaluation indexes of aerospace embedded real-time Linux, select the risk sets and establish the evaluation index system and evaluation set of aerospace embedded real-time Linux.

2) Obtain the weight of each risk factor by the method of expert scoring; the weight coefficient matrix is generated by the inverse cloud generator.

3) Refer to the risk ranking table. In order to reduce the impact of fuzziness and subjectivity, the evaluation value is obtained by combining the measured data with expert scoring, and then the comprehensive evaluation matrix is generated by the inverse cloud generator.

4) The digital features of the evaluation cloud model are obtained by weight coefficient matrix and comprehensive evaluation matrix. The forward cloud generator is used to generate the “cloud droplet” diagrams of the evaluation cloud and the evaluation cloud model; the risk assessment level of the space embedded Linux is obtained.

The flow chart of risk assessment based on cloud model is shown in Figure2.

Figure 2

New window| Download| PPT slide

Figure 2. Flow chart of risk assessment

3.1. Establishment of Risk Evaluation System

Because of the complexity of the operating system requirements, it is difficult to make a systematic, scientific and comprehensive evaluation of the space embedded operating system only by real-time indicators. So, it is necessary to establish a complete evaluation system in the control system of manned spaceflight. This paper combines the quality model and basic measurement and testing points of military software products defined in GJB5236-2004 Military Software Quality Measurement Standard and GJB7706-2012 Military Embedded Operating System Evaluation Requirements. At the same time, according to the special requirements of embedded real-time operating system in aerospace field, the evaluation index of aerospace embedded real-time Linux is customized.

This paper refers to the real-time performance test indicators of Rhealstone and Hartstone real-time operating systems. The real-time performance indicators are further refined into four indicators: interrupt delay time, task preemption time, task switching time and semaphore shuffling time. The deadlock release time and system throughput in Rhealstone method were ignored, because priority inheritance has been used to avoid the problem of priority inversion in real-time patches such as Xenomai, RTAI and Preempt-RT[14]. At the same time, for space embedded real-time systems, the importance of real-time is far greater than the overall throughput of the system. In fact, system throughput is an important indicator to measure the server, which is not very significant for space embedded systems.

Following the principles of objectivity, testability, completeness, independence, consistency and simplicity in selecting evaluation indicators[15], and considering the special requirements for real-time, reliability and compatibility in the field of manned spaceflight, this paper selects interrupt delay time, task preemption time, task switching time, semaphore shuffling time, kernel stability, hardware compatibility and transplant complexity to comprehensively evaluate the space embedded real-time Linux operating system.

The above seven indicators cover threemajor aspects of the embedded Linux operating system. Interrupt delay time, task preemption time, task switching time and semaphore shuffling time are part ofreal-time category; kernel stability belongs to security category, hardware compatibility and transplant complexity belong to the compatibility category.

Firstly, to describe the risk level, an embedded Linux comment set is given by the cloud model. Each comment has a boundary [A_min, A_max], and the boundary is a bilateral boundary. The numerical characteristics of each comment cloud model can be approximated by Formula (3)[16].

(3)

\{\begin{matrix} E_{x} = (A_{\max} + A_{\min}) / 2 \\ E_{n} = (A_{\max} - A_{\min}) / 6 \\ H_{e} = m \end{matrix}

A_max and A_min are the upper and lower boundaries of a risk rating respectively;m is constant and is usually adjusted according to the degree of fuzziness of risk concentration[17].

Considering the tolerance of space embedded Linux system to various risks, different risk levels are given evaluation scope and rating, as shown in Table 1.

Table 1. Risk assessment grade

Grade	1	2	3	4	5
Risk	Risk-free	Low risk	Medium risk	High risk	Unacceptable
Evaluation	[0,2]	[2,4]	[4,6]	[6,8]	[8,10]

New window| CSV

From Table 1, five risk models are represented by cloud models to construct membership functions. Risk-free (grade 1) means this kind of Linux can be usedin directly; low-risk(grade 2) means this kind of Linux can be used in some specified scenario; medium-risk(grade 3) means this kind of Linux may be used after modified; high-risk(grade 4) or unacceptable(grade 5) means this kind of Linux is not applicable to manned spaceflight.

According to the classification of risk levels and Formula (1), the risk assessment cloud model of space embedded Linux is shown in Table 2.

Table 2. Cloud model of risk assessment

Risk assessment grade	Parameters (E_x, E_n, H_e)
1	(1,1,0.2)
2	(3, 1,0.2)
3	(5, 1,0.2)
4	(7, 1,0.2)
5	(9, 1,0.2)

New window| CSV

3.2. Calculation of Weight Coefficient Matrix and Comprehensive Evaluation Matrix

After obtaining the risk set of space embedded Linux, the corresponding weight set and evaluation matrix are A=[a₁,a₂,…,a_n], R=[r₁,r₂,…,r_n], and n is the number of the risk factors. In this paper, cloud model is used to replace the traditional membership function to calculate the weight coefficient matrix and comprehensive evaluation matrix. The expected value, entropy and super entropy of expert scoring model cloud model can be calculated by inverse cloud generator. The weight matrix and the comprehensive evaluation matrix are as follows:

(4)A=[a₁,a₂,…,a_n]=

{[\begin{matrix} E_{x_{a 1}} & \begin{matrix} E_{n_{a 1}} & H_{e_{a 1}} \end{matrix} \\ \begin{matrix} E_{x_{a 2}} \\ ⋮ \\ E_{x_{an}} \end{matrix} & \begin{matrix} \begin{matrix} E_{n_{a 2}} \\ ⋮ \\ E_{n_{an}} \end{matrix} & \begin{matrix} H_{e_{a 2}} \\ ⋮ \\ H_{e_{an}} \end{matrix} \end{matrix} \end{matrix}]}^{T}

(5)R=[r₁,r₂,…,r_n]^T=

{[\begin{matrix} E_{x_{1}} & \begin{matrix} E_{n_{1}} & H_{e_{1}} \end{matrix} \\ \begin{matrix} E_{x_{2}} \\ ⋮ \\ E_{x_{n}} \end{matrix} & \begin{matrix} \begin{matrix} E_{n_{2}} \\ ⋮ \\ E_{n_{n}} \end{matrix} & \begin{matrix} H_{e_{2}} \\ ⋮ \\ H_{e_{n}} \end{matrix} \end{matrix} \end{matrix}]}^{T}

According to the traditional expert scoring method, each value of Formula (5) has fuzziness and subjectivity. This paper adopts the method of combining expert scoring with measured data, greatly reducing the fuzziness and subjectivity of the comprehensive evaluation matrix while improving the credibility of the evaluation results.

At last, the digital signature B of aerospace embedded Linux risk assessment cloud model is obtained by

(6)B = A°R = (E_x, E_n, H_e)

In Formula (6), “ ° ” is the operator of synthetic computation. It is also a symbol of cloud computing. The operation rules of cloud computing are shown in Table 3.

Table 3. Algorithm of cloud

Symbol	$E_{x}$	$E_{n}$	$H_{e}$
+	$E_{x_{1}}$ + $E_{x_{2}}$	$\sqrt[]{E_{n_{1}}^{2} + E_{n_{2}}^{2}}$	$\sqrt[]{H_{e_{1}}^{2} + H_{e_{2}}^{2}}$
-	$E_{x_{1}}$ - $E_{x_{2}}$	$\sqrt[]{E_{n_{1}}^{2} + E_{n_{2}}^{2}}$	$\sqrt[]{H_{e_{1}}^{2} + H_{e_{2}}^{2}}$
$\times$	$E_{x_{1}}$  $E_{x_{2}}$	$\|E_{x_{1}} E_{x_{2}}\| \sqrt[]{{(\frac{E_{n_{1}}}{E_{x_{1}}})}^{2} + {(\frac{E_{n_{2}}}{E_{x_{2}}})}^{2}}$	$\|E_{x_{1}} E_{x_{2}}\| \sqrt[]{{(\frac{H_{e_{1}}}{E_{x_{1}}})}^{2} + {(\frac{H_{e_{2}}}{E_{x_{2}}})}^{2}}$
÷	$E_{x_{1}}$ ÷ $E_{x_{2}}$	$\|\frac{E_{x_{1}}}{E_{x_{2}}}\| \sqrt[]{{(\frac{E_{n_{1}}}{E_{x_{1}}})}^{2} + {(\frac{E_{n_{2}}}{E_{x_{2}}})}^{2}}$	$\|\frac{E_{x_{1}}}{E_{x_{2}}}\| \sqrt[]{{(\frac{H_{e_{1}}}{E_{x_{1}}})}^{2} + {(\frac{H_{e_{2}}}{E_{x_{2}}})}^{2}}$

New window| CSV

The comprehensive evaluation results are calculated according to Table 3.

(7)

E_{x}

E_{x_{a 1}}



E_{x_{1}}

E_{x_{a 2}}



E_{x_{2}}

+…+

E_{x_{an}}



E_{x_{n}}

(8)

E_{n}

= {|

E_{x_{a 1}} E_{x_{1}}

[(

E_{n_{a 1}}

E_{x_{a 1}}

)²+ (

E_{n_{1}}

E_{x_{1}}

)²]^1/2|²+ |

E_{x_{a 2}} E_{x_{2}}

[(

E_{n_{a 2}}

E_{x_{a 2}}

)²+ (

E_{n_{2}}

E_{x_{2}}

)²]^1/2|²+…+ |

E_{x_{an}} E_{x_{n}}

[(

E_{n_{an}}

E_{x_{an}}

)²+ (

E_{n_{n}}

E_{x_{n}}

)²]^1/2|²}^1/2

(9)He={|

E_{x_{a 1}} E_{x_{1}}

[(

H_{e_{a 1}}

E_{x_{a 1}}

)²+ (

H_{e_{1}}

E_{x_{1}}

)²]^1/2|²+ |

E_{x_{a 2}} E_{x_{2}}

[(

H_{e_{a 2}}

E_{x_{a 2}}

)²+ (

H_{e_{2}}

E_{x_{2}}

)²]^1/2|²+…+|

E_{x_{an}} E_{x_{n}}

[(

H_{e_{an}}

E_{x_{an}}

)²+ (

H_{e_{n}}

E_{x_{n}}

)²]^1/2|²}^1/2

4. Analysis of Engineering Case

In the design of manned spaceflight application system, it is inevitable to encounter the problem of risk assessment and selection of load control system. Considering the practical problems of risk assessment and selection of space embedded Linux operating system in manned spaceflight project, this paper uses cloud-based risk assessment model to evaluate embedded Linux system and its real-time derivatives, and gives the selection scheme according to the evaluation results.

4.1. Application of Risk Assessment based on Cloud Model

In this paper, a risk assessment model of space embedded Linux and its real-time derivatives based on cloud model has been proposed. Linux based on PREEMPT-RT, Xenomai, RTAI patch and Linux std.are taken as an example to validate the risk assessment based on cloud model in this paper. The validity of the model is verified by the following example.

In order to minimize the impact of human subjective factors, the method of expert scoring is adopted in the comprehensive evaluation of space embedded Linux system. First, the seven risk factors selected should be weightedbecause each of them has different impact on the space embedded Linux system.

According to the importance of each risk factor, each risk factor is scored by experts; the corresponding weight set of each risk is obtained. Then, the inverse cloud generator is used to get the weight cloud of each risk. The weight coefficient Ais calculated according to Formula (4). In different application scenarios, the weight coefficient matrix may be different. This article takes the application scenario focusing on system control as an example. The weight coefficient matrix Ais shown below.

A=[a₁,a₂,…,a_n]= ${[\begin{matrix} 0.25 & \begin{matrix} 0.072 & 0.034 \end{matrix} \\ \begin{matrix} 0.20 \\ 0.18 \\ \begin{matrix} 0.16 \\ 0.12 \\ \begin{matrix} 0.10 \\ 0.08 \end{matrix} \end{matrix} \end{matrix} & \begin{matrix} \begin{matrix} 0.086 \\ 0.068 \\ \begin{matrix} 0.106 \\ 0.058 \\ \begin{matrix} 0.086 \\ 0.062 \end{matrix} \end{matrix} \end{matrix} & \begin{matrix} 0.018 \\ 0.046 \\ \begin{matrix} 0.072 \\ 0.016 \\ \begin{matrix} 0.025 \\ 0.016 \end{matrix} \end{matrix} \end{matrix} \end{matrix} \end{matrix}]}^{T}$

Combining with the risk assessment criteria of space embedded Linux, for interrupt delay time, task preemption time, task switching time and semaphore shuffling time and other quantitative factors, the measured value method is used. Expert scoring is used for qualitative factors such as kernel stability, hardware compatibility and transplant complexity. Then, the comprehensive evaluation matrix Ris obtained according to Formula (4). The comprehensive evaluation matrix R_p, R_x, R_r, and R_sbased on PREEMPT-RT, Xenomai, RTAI patch and Linux std. are shown below.

R_p=[r₁,r₂,…,r_n]^T= $[\begin{matrix} 3.04 & \begin{matrix} 0.156 & 0.134 \end{matrix} \\ \begin{matrix} 2.78 \\ 2.35 \\ \begin{matrix} 2.02 \\ 1.85 \\ \begin{matrix} 1.01 \\ 1.14 \end{matrix} \end{matrix} \end{matrix} & \begin{matrix} \begin{matrix} 0.086 \\ 0.268 \\ \begin{matrix} 0.106 \\ 0.158 \\ \begin{matrix} 0.286 \\ 0.162 \end{matrix} \end{matrix} \end{matrix} & \begin{matrix} 0.118 \\ 0.066 \\ \begin{matrix} 0.172 \\ 0.116 \\ \begin{matrix} 0.127 \\ 0.011 \end{matrix} \end{matrix} \end{matrix} \end{matrix} \end{matrix}]$

R_x=[r₁,r₂,…,r_n]^T= $[\begin{matrix} 2.24 & \begin{matrix} 0.125 & 0.134 \end{matrix} \\ \begin{matrix} 2.34 \\ 2.46 \\ \begin{matrix} 2.62 \\ 4.85 \\ \begin{matrix} 5.41 \\ 6.24 \end{matrix} \end{matrix} \end{matrix} & \begin{matrix} \begin{matrix} 0.324 \\ 0.321 \\ \begin{matrix} 0.206 \\ 0.238 \\ \begin{matrix} 0.186 \\ 0.262 \end{matrix} \end{matrix} \end{matrix} & \begin{matrix} 0.076 \\ 0.106 \\ \begin{matrix} 0.122 \\ 0.216 \\ \begin{matrix} 0.267 \\ 0.021 \end{matrix} \end{matrix} \end{matrix} \end{matrix} \end{matrix}]$

R_r=[r₁,r₂,…,r_n]^T= $[\begin{matrix} 1.64 & \begin{matrix} 0.256 & 0.134 \end{matrix} \\ \begin{matrix} 2.08 \\ 2.15 \\ \begin{matrix} 2.22 \\ 4.85 \\ \begin{matrix} 6.21 \\ 6.84 \end{matrix} \end{matrix} \end{matrix} & \begin{matrix} \begin{matrix} 0.186 \\ 0.064 \\ \begin{matrix} 0.106 \\ 0.258 \\ \begin{matrix} 0.106 \\ 0.162 \end{matrix} \end{matrix} \end{matrix} & \begin{matrix} 0.108 \\ 0.106 \\ \begin{matrix} 0.112 \\ 0.016 \\ \begin{matrix} 0.103 \\ 0.013 \end{matrix} \end{matrix} \end{matrix} \end{matrix} \end{matrix}]$

R_s=[r₁,r₂,…,r_n]^T= $[\begin{matrix} 6.84 & \begin{matrix} 0.196 & 0.034 \end{matrix} \\ \begin{matrix} 5.91 \\ 5.17 \\ \begin{matrix} 4.71 \\ 1.65 \\ \begin{matrix} 1.01 \\ 1.20 \end{matrix} \end{matrix} \end{matrix} & \begin{matrix} \begin{matrix} 0.173 \\ 0.131 \\ \begin{matrix} 0.139 \\ 0.0558 \\ \begin{matrix} 0.029 \\ 0.162 \end{matrix} \end{matrix} \end{matrix} & \begin{matrix} 0.218 \\ 0.066 \\ \begin{matrix} 0.172 \\ 0.116 \\ \begin{matrix} 0.027 \\ 0.011 \end{matrix} \end{matrix} \end{matrix} \end{matrix} \end{matrix}]$

Now we have got the weight coefficient matrix A and the comprehensive evaluation matrix R. The numerical characteristics of the cloud model of the final evaluation result are obtained through Formula (6).

B_p=A°R_p=[2.4764,0.792523761,0.296632107]

B_x=A°R_x= [3.5122,0.792523761, 0.296632107]

B_r=A°R_r= [3.3184,0.820989366, 0.285042379]

B_s=A°R_s= [4.9712,0.820989366, 0.285042379]

4.2. Test Results and Analysis

Cloud generator was used to show the five evaluation grades and the final result of evaluation by MATLAB; the charts are as shown in Figure 3. It is not difficult to see from Figure 3(a) to Figure 3(d) that the evaluation cloud—the black *, for space embedded Linux systems based on PREEMPT-RT patch, is mostly in the “low-risk” comment cloud with a small part in the “risk-free”. Space embedded Linux systems based on Xenomai and RTAI patch coincide with the “low-risk” while Linux std. is between “low-risk” and “medium-risk”.

Figure 3

New window| Download| PPT slide

Figure 3. (a) Final result of Linux based on Preempt-rt; (b) Final result of Linux based on Xenomai; (c) Final result of Linux based on RTAI; (d) Final result of Linuxstd

There is no doubt that Linux system based on PREEMPT-RT patch is the best choice among the four alternatives. It is consistent with the actual situation of Linux system based on PREEMPT_RT patch. Because Linux based on PREEMPT-RT patch is a single kernel, it has good real-time performance and portability. It also verifies the effectiveness of cloud based embedded Linux risk assessment model.

5. Conclusion

The risk assessment result of space embedded Linux based on cloud model conforms to the real investigation situation, it provides a scientific basis for the evaluation and selection of embedded Linux system for manned space engineering, and can reduce the impact of subjective judgment to a great extent. The cloud model can not only fully reflect the fuzziness of the risk assessment of space embedded Linux, but also make the assessment results more intuitive. This makes the assessment results more objective, effective and credible. This method also provides a train of thought for further expansion of cloud models.

Reference

By original order

By published year

By cited times in this paper

By Journal Impact factor

[1]

Y. S.

Liu

, Y.

, J. B.

Wang

, and C. Y.

, “

Research on Real Time Performance Optimization of Space Embedded Linux

,” Aerospace Control, Vol. 3 No. 10, pp. 57-62, 78, March 2018