A Cloud-Computing-Based Portable Networked Ground Station System for Microsatellites

Microsatellites have attracted a large number of scholars and engineers because of their portability and distribution characteristics. The ground station suitable for microsatellite service has become an important research topic. In this paper, we propose a networked ground station and verify it on our own microsatellite. The specific networked ground station system consists of multiple ground nodes. They can work together to complete data transmission tasks with higher efficiency. After describing our microsatellite project, a reasonable distribution of ground nodes is given. A cloud computing model is used to realize the coordination of multiple ground nodes. An adaptive communication system between satellites and ground stations is used to increase link efficiency. Extensive on-orbit experiments were used to validate our design. The experimental results show that our networked ground station has excellent performance in data transmission capability. Finally, the specific cloud-computing-based ground station network successfully completes our satellite mission.


Introduction
With the rapid development of satellite technology, a large number of new topics are proposed every year [1][2][3]. Among the many research directions, the topic of microsatellites has attracted the attention of a large number of scholars and engineers. Its advantages include miniaturization, low cost and short development cycle. In order to realize a superior microsatellite system, a high-performance data transmission system is essential.
In addition to the design of satellites, the ground station is an indispensable component of satellite systems [4]. The harsh space environment, low power ability and small size of antennas make the communication ability of microsatellites poor [5]. Traditional ground stations cannot meet the stringent requirements of microsatellites For microsatellites, a sun-synchronous orbit (SSO) is utilized for more energy from the sun. Unfortunately, the distance from the SSO to a fixed ground node could change three times with the movement of satellites [6], which will affect the received signal strength. The traditional solution for this inherent flaw is to reserve more margin in the link budget, but it is a luxury and sometimes impossible for microsatellites. For low transmission delay, a low altitude is selected by the SSO. However, on the other hand, the lower the orbit altitude, the shorter the link duration. This drawback makes information throughput insufficient for telemeter (TM) and telecommand (TC).
Recently, there have been many studies on satellite ground stations. Refs. [7][8][9] propose some hardware designs of ground station equipment are introduced. Novel antenna and GPU boards are found to improve ground station performance at the cost of resource consumption. In [10], a throughput-based transmission plan is introduced. In [11], Liashkevich designs an X-band university ground station with the SDR platform

Related Work
There are many related satellite ground station studies. Ref. [16] proposes a communitydriven approach used to solve the access problem between satellites and ground stations. Although this approach is based on cloud servers, it is unable to simultaneously control multiple ground stations. The cloud server only acts as a database. The link duration is not enough to transmit massive data. Refs. [21,22] proposes a networked ground station which is introduced to realize data transmission. However, the management of data does not include task assignment and resource management. This defect greatly reduces the efficiency of data reception. Ref. [23] proposes a ground station assignment strategy which is used to increase communication efficiency. However, a lot of computation increases the size and power consumption of the ground station, which cannot satisfy the miniaturization requirements. Ref. [24] proposes an adaptive feeder link which is used to realize the adaptivity of communication systems. However, this optical system cannot be used in our satellites. Refs. [25][26][27][28] propose many adaptive systems for mobile communication, intervehicle communication and laser communication. Due to the dynamic changes in satellite communication, they cannot be used directly in satellite-to-ground communication systems. Refs. [29][30][31] propose many cloud computing systems for satellite application. They contain a large number of data interaction strategies. However, they cannot adequately balance performance with efficiency.
To further understand our requirements, the ten most relevant studies were compared in detail. As shown in Table 1, there are four important parameters defined as judging criteria. Most existing research only meets one requirement. This will greatly hinder the execution of our SSS-2A mission. Table 1. System requirements analysis. ( √ means meeting the requirements, means not).
The main innovations of this paper are as follows. In order to improve the link duration time, a virtual link is constructed by combining multiple ground nodes. This function can only be realized based on cloud computing. In order to increase communication efficiency, adaptive communication based on a feedback system is constructed. To meet the requirements of miniaturization and low power consumption, a ground station based on the SDR architecture is designed. This system dynamically controls the power supply of all devices according to cloud commands. This measure greatly reduces power consumption. An on-orbit verification experiment is more convincing than other simulations and tests.

Mission Description
The SSS project organized by APSCO contains three Student Small Satellite subprojects: SSS-1, SSS-1A and SSS-2A. SSS-1 is a small satellite weighing about 50 kg, carrying an ADS-B receiver as its main payload. SSS-1A and SSS-2A are microsatellites with 3U volume. SSS-2A, equipped with an AIS receiver and intersatellite communication as payloads, was developed by Shanghai Jiao Tong University (SJTU). The UHF and VHF bands are selected as the main channels for telemeter (TM), telecommand (TC) and data transmission (DT). In addition to the development of microsatellites, the ground station system is also developed by ourselves.

Orbit Design
A sun-synchronous orbit (SSO), also called heliosynchronous orbit, is the near-polar orbit around a planet. In SSO, the satellite passes over any given point of the planet's surface at the same local mean solar time. It always maintains the same relationship with the sun. This advantage is beneficial to small satellites because of their low power consumption.
In the SSS project, a 500 km SSO is defined as the operating orbit. The distance from the satellite to a given ground node changes from 500 km to 1700 km, and the pitch angle changes from 90 • to 10 • . The relative velocity changes from 0 to 7 km/s. As shown in Figure 1, the satellite moves clockwise around Earth, from Points A to B and C. The space attenuation of the link changes from −130 dB to −160 dB.
consumption. An on-orbit verification experiment is more convincing than other simulations and tests. The SSS project organized by APSCO contains three Student Small Satellite subprojects: SSS-1, SSS-1A and SSS-2A. SSS-1 is a small satellite weighing about 50 kg, carrying  an ADS-B receiver as its main payload. SSS-1A and SSS-2A are microsatellites with 3U  volume. SSS-2A, equipped with an AIS receiver and intersatellite communication as payloads, was developed by Shanghai Jiao Tong University (SJTU). The UHF and VHF bands are selected as the main channels for telemeter (TM), telecommand (TC) and data transmission (DT). In addition to the development of microsatellites, the ground station system is also developed by ourselves.

Orbit Design
A sun-synchronous orbit (SSO), also called heliosynchronous orbit, is the near-polar orbit around a planet. In SSO, the satellite passes over any given point of the planet's surface at the same local mean solar time. It always maintains the same relationship with the sun. This advantage is beneficial to small satellites because of their low power consumption.
In the SSS project, a 500 km SSO is defined as the operating orbit. The distance from the satellite to a given ground node changes from 500 km to 1700 km, and the pitch angle changes from 90° to 10°. The relative velocity changes from 0 to 7 km/s. As shown in Figure 1, the satellite moves clockwise around Earth, from Points A to B and C. The space attenuation of the link changes from −130 dB to −160 dB. For a definite orbit, the link attenuation can be calculated in detail. The minimum received power is −160 dBm. The maximum doppler offset is 10 KHz for the UHF band and 6 KHz for the VHF band. The results of these calculations are important boundary conditions for the design of ground stations. For a definite orbit, the link attenuation can be calculated in detail. The minimum received power is −160 dBm. The maximum doppler offset is 10 KHz for the UHF band and 6 KHz for the VHF band. The results of these calculations are important boundary conditions for the design of ground stations.

TM, TC and DT Analysis
For the requirements of SSS-2A, there are three different data packets in our communication system: telecommand (TC), telemeter (TM) and data transmission (DT). The TC packet contains 128 bits and can be divided into four components. The TM and DT packets contain 2048 bits and can also be divided into four components. The details of the three different packets are shown in Figure 2.

TM, TC and DT Analysis
For the requirements of SSS-2A, there are three different data packets in our communication system: telecommand (TC), telemeter (TM) and data transmission (DT). The TC packet contains 128 bits and can be divided into four components. The TM and DT packets contain 2048 bits and can also be divided into four components. The details of the three different packets are shown in Figure 2. According to the agreements, there are two different modulation methods that can back each other up, namely GMSK and BPSK. Two different data rates, 9.6 kbps and 4.8 kbps, are available. Two different coding types, RS coding and convolution coding, are options for our design. As shown in Figure 3, the transmit interval between two packets is also optional. TM packets should be responsible for the transfer of the data from the integrated electronic subsystem, attitude and orbit subsystem and power subsystem. Since a large number of data packets will be automatically generated by the satellite, it is very important to efficiently arrange the data transmission. TC packets should be responsible for the transfer of the data from the ground command center. DT packets should be responsible for the transfer of the data from payloads.
According to ITU standards, the VHF band is selected as the TC channel, and the UHF band is selected as the TM and DT channel. These bands are also amateur radio frequency bands; thus, an amateur license is also a legal permit for our ground stations.

Microsatellite Communication System Design
Since SSS-2A is a microsatellite, the size of the motherboards of all equipment must meet the conditions. As shown in Figure 4, the envelopes of the UVH transceiver and UVH antenna are compatible with the PC/104 mechanical specifications. According to the agreements, there are two different modulation methods that can back each other up, namely GMSK and BPSK. Two different data rates, 9.6 kbps and 4.8 kbps, are available. Two different coding types, RS coding and convolution coding, are options for our design. As shown in Figure 3, the transmit interval between two packets is also optional.

TM, TC and DT Analysis
For the requirements of SSS-2A, there are three different data packets in our communication system: telecommand (TC), telemeter (TM) and data transmission (DT). The TC packet contains 128 bits and can be divided into four components. The TM and DT packets contain 2048 bits and can also be divided into four components. The details of the three different packets are shown in Figure 2. According to the agreements, there are two different modulation methods that can back each other up, namely GMSK and BPSK. Two different data rates, 9.6 kbps and 4.8 kbps, are available. Two different coding types, RS coding and convolution coding, are options for our design. As shown in Figure 3, the transmit interval between two packets is also optional. TM packets should be responsible for the transfer of the data from the integrated electronic subsystem, attitude and orbit subsystem and power subsystem. Since a large number of data packets will be automatically generated by the satellite, it is very important to efficiently arrange the data transmission. TC packets should be responsible for the transfer of the data from the ground command center. DT packets should be responsible for the transfer of the data from payloads.
According to ITU standards, the VHF band is selected as the TC channel, and the UHF band is selected as the TM and DT channel. These bands are also amateur radio frequency bands; thus, an amateur license is also a legal permit for our ground stations.

Microsatellite Communication System Design
Since SSS-2A is a microsatellite, the size of the motherboards of all equipment must meet the conditions. As shown in Figure 4, the envelopes of the UVH transceiver and UVH antenna are compatible with the PC/104 mechanical specifications. TM packets should be responsible for the transfer of the data from the integrated electronic subsystem, attitude and orbit subsystem and power subsystem. Since a large number of data packets will be automatically generated by the satellite, it is very important to efficiently arrange the data transmission. TC packets should be responsible for the transfer of the data from the ground command center. DT packets should be responsible for the transfer of the data from payloads.
According to ITU standards, the VHF band is selected as the TC channel, and the UHF band is selected as the TM and DT channel. These bands are also amateur radio frequency bands; thus, an amateur license is also a legal permit for our ground stations.

Microsatellite Communication System Design
Since SSS-2A is a microsatellite, the size of the motherboards of all equipment must meet the conditions. As shown in Figure 4, the envelopes of the UVH transceiver and UVH antenna are compatible with the PC/104 mechanical specifications.
The UVH transceiver consists of digital baseband components and RF frontend components. It can also transmit and receive signals. In our mission, the downlink frequency for TM and DT is 435.775 MHz, and the uplink frequency for TC is 145.985 MHz. The internal power amplifier (PA) outputs a 0.5 W RF signal to the antenna. The average gain of the specific antenna is −3 dB; thus, the equivalent isotropically radiated power (EIRP) can be defined as 24 dBm.  The UVH transceiver consists of digital baseband components and RF frontend components. It can also transmit and receive signals. In our mission, the downlink frequency for TM and DT is 435.775 MHz, and the uplink frequency for TC is 145.985 MHz. The internal power amplifier (PA) outputs a 0.5 W RF signal to the antenna. The average gain of the specific antenna is −3 dB; thus, the equivalent isotropically radiated power (EIRP) can be defined as 24 dBm.
For miniaturization, a set of dipole antennas is placed on the top of the satellite. Two orthogonal linear polarization characteristics can increase the isolation of transceivers and also increase the difficulty of ground equipment.

Ground Station Network
Short link duration is the fatal flaw of LEO microsatellites. Fortunately, this shortcoming can be compensated for with a collaboration method. Multiple nodes work together to serve the same satellite to increase the total link duration.
Before the design of ground stations, suitable locations should be selected. Since urban noise, building occlusion and environmental factors can all affect the performance of the ground stations, the location selection of the ground stations is particularly important. As an important test and experiment, sparsely populated places are more suitable for building ground stations. However, the specific location distribution of the ground stations also depends on the mission requirements.

Network Topology
For excellent payload performance, a large amount of data needs to be transmitted from the satellite. To compensate for the short link duration caused by orbit characteristics, multiple ground nodes should be scattered throughout the country. They should be spread evenly across longitudes and latitudes. Based on the actual situation in China, six regional ground stations are selected.  For miniaturization, a set of dipole antennas is placed on the top of the satellite. Two orthogonal linear polarization characteristics can increase the isolation of transceivers and also increase the difficulty of ground equipment.

Ground Station Network
Short link duration is the fatal flaw of LEO microsatellites. Fortunately, this shortcoming can be compensated for with a collaboration method. Multiple nodes work together to serve the same satellite to increase the total link duration.
Before the design of ground stations, suitable locations should be selected. Since urban noise, building occlusion and environmental factors can all affect the performance of the ground stations, the location selection of the ground stations is particularly important. As an important test and experiment, sparsely populated places are more suitable for building ground stations. However, the specific location distribution of the ground stations also depends on the mission requirements.

Network Topology
For excellent payload performance, a large amount of data needs to be transmitted from the satellite. To compensate for the short link duration caused by orbit characteristics, multiple ground nodes should be scattered throughout the country. They should be spread evenly across longitudes and latitudes. Based on the actual situation in China, six regional ground stations are selected.  Figure 5.
The coverage of the selected six ground nodes contains most of the country's land area. The specific topology can guarantee more connection opportunities. In order to verify the topology, we simulate the link access between the satellite and ground nodes with the commercial software STK. As shown in Table 2, the proposed topology obtains 18 connection opportunities in 24 h. Each link has a different duration, pitch angle and distance. For instance, Links 5, 8, 12 and 15-17 have a duration long enough to transmit large data packets. Different tasks should be assigned to different links. The coverage of the selected six ground nodes contains most of the country's land area. The specific topology can guarantee more connection opportunities. In order to verify the topology, we simulate the link access between the satellite and ground nodes with the commercial software STK. As shown in Table 2, the proposed topology obtains 18 connection opportunities in 24 h. Each link has a different duration, pitch angle and distance. For instance, Links 5, 8, 12 and 15-17 have a duration long enough to transmit large data packets. Different tasks should be assigned to different links. For some links, such as Links 4 and 5, their overlapping parts need to be further processed for higher efficiency. Some adjacent connections can be constructed as a continuous  For some links, such as Links 4 and 5, their overlapping parts need to be further processed for higher efficiency. Some adjacent connections can be constructed as a continuous link. This virtual link consists of links of multiple different ground nodes. By these methods, it is possible to artificially establish longer durations.
As described in Table 2 and shown in Figure 6, Links 3-6 overlap each other. When selecting the link from Links 3 to 4 and Links 5 to 6 in turn, a virtual continuous link duration from 10:41:54 to 10:54:42 can be established. This virtual link duration lasts 720 s, which is far longer than any single link duration. Similarly, Links 12, 13 and 15 can establish another virtual link duration. Table 2 and shown in Figure 6, Links 3-6 overlap each other. When selecting the link from Links 3 to 4 and Links 5 to 6 in turn, a virtual continuous link duration from 10:41:54 to 10:54:42 can be established. This virtual link duration lasts 720 s, which is far longer than any single link duration. Similarly, Links 12, 13 and 15 can establish another virtual link duration. Through collaboration of the virtual link and multiple ground stations, the duration of the new ground link is greatly increased. Compared with the individual mode, collaboration can save a lot of overhand between satellites and ground stations.

As described in
In this section, we proposed a ground station topology with six ground nodes and simulate it with STK. This appropriate topology can improve the link duration and coverage of ground stations. However, cooperative control, establishment of virtual links and reconfiguration of overlapping links all require a large amount of computation. This computing power greatly increases the burden on the ground station. Fortunately, cloud computing technology can solve this problem.

Cloud Computing
For excellent satellite-to-ground data transmission capability, multiple ground stations are required to collaborate. It is also an important task to collect received data packets scattered at various ground stations. Predicting the link access according to the TLE of satellites and allocating access tasks to each ground station will also consume a lot of computing power. These main functions all rely on a series of self-developed algorithms. They will be running on cloud servers.

Massive Data Management
In each connection opportunity, hundreds of TM and DT packets can be received from SSS-2A, and dozens of TC packets can also be transmitted. A large amount of data processing is complex. The processing of virtual links is more complicated.
As shown in Figure 7, through the cloud server, six scattered ground stations can be managed uniformly. The cloud data center will collect and store the data packets in real Through collaboration of the virtual link and multiple ground stations, the duration of the new ground link is greatly increased. Compared with the individual mode, collaboration can save a lot of overhand between satellites and ground stations.
In this section, we proposed a ground station topology with six ground nodes and simulate it with STK. This appropriate topology can improve the link duration and coverage of ground stations. However, cooperative control, establishment of virtual links and reconfiguration of overlapping links all require a large amount of computation. This computing power greatly increases the burden on the ground station. Fortunately, cloud computing technology can solve this problem.

Cloud Computing
For excellent satellite-to-ground data transmission capability, multiple ground stations are required to collaborate. It is also an important task to collect received data packets scattered at various ground stations. Predicting the link access according to the TLE of satellites and allocating access tasks to each ground station will also consume a lot of computing power. These main functions all rely on a series of self-developed algorithms. They will be running on cloud servers.

Massive Data Management
In each connection opportunity, hundreds of TM and DT packets can be received from SSS-2A, and dozens of TC packets can also be transmitted. A large amount of data processing is complex. The processing of virtual links is more complicated.
As shown in Figure 7, through the cloud server, six scattered ground stations can be managed uniformly. The cloud data center will collect and store the data packets in real time. Similarly, the TC packet will also be distributed to the corresponding ground station. In the cloud server, there are three different databases for storing and processing different data packets. These data can be visited by an authorized customer via a network interface. time. Similarly, the TC packet will also be distributed to the corresponding ground station. In the cloud server, there are three different databases for storing and processing different data packets. These data can be visited by an authorized customer via a network interface.

Upload Data Flow
Once a TM or DT packet is received successfully, a set of raw data with 256 bytes is stored in the ground station. It will be uploaded to the cloud data center by the terrestrial network immediately. As shown in Figure 8, the cloud server will store raw data in data-base1. The raw data are unable to express the real information of satellites or payloads. A data analysis block is designed to translate raw data into numerical data, which is identifiable by the customer. The numerical data are stored in database 2. Any authorized user can access the desired information through the UI interface. There is a feedback mechanism to ensure the reliability of data collection. The ground station will keep the uploaded data until it receives an acknowledgment (ACK) from the cloud server.

Download Data Flow
The download flow is responsible for TC packets and satellite update packets. As shown in Figure 9, an authorized user can transmit data packets to specific ground stations. Similarly, TC packets should also be converted.

Upload Data Flow
Once a TM or DT packet is received successfully, a set of raw data with 256 bytes is stored in the ground station. It will be uploaded to the cloud data center by the terrestrial network immediately. As shown in Figure 8, the cloud server will store raw data in database1. The raw data are unable to express the real information of satellites or payloads. A data analysis block is designed to translate raw data into numerical data, which is identifiable by the customer. The numerical data are stored in database 2. Any authorized user can access the desired information through the UI interface. time. Similarly, the TC packet will also be distributed to the corresponding ground station.
In the cloud server, there are three different databases for storing and processing different data packets. These data can be visited by an authorized customer via a network interface.

Upload Data Flow
Once a TM or DT packet is received successfully, a set of raw data with 256 bytes is stored in the ground station. It will be uploaded to the cloud data center by the terrestrial network immediately. As shown in Figure 8, the cloud server will store raw data in data-base1. The raw data are unable to express the real information of satellites or payloads. A data analysis block is designed to translate raw data into numerical data, which is identifiable by the customer. The numerical data are stored in database 2. Any authorized user can access the desired information through the UI interface. There is a feedback mechanism to ensure the reliability of data collection. The ground station will keep the uploaded data until it receives an acknowledgment (ACK) from the cloud server.

Download Data Flow
The download flow is responsible for TC packets and satellite update packets. As shown in Figure 9, an authorized user can transmit data packets to specific ground stations. Similarly, TC packets should also be converted. There is a feedback mechanism to ensure the reliability of data collection. The ground station will keep the uploaded data until it receives an acknowledgment (ACK) from the cloud server.

Download Data Flow
The download flow is responsible for TC packets and satellite update packets. As shown in Figure 9, an authorized user can transmit data packets to specific ground stations. Similarly, TC packets should also be converted. Any data processed by the cloud data center will be stored for a long time. In the case of negligent operation, any transmitted data packets will be recorded in a log. For excellent performance, different software will be applied in our cloud data center, as shown in Table 3. For miniaturization and low power consumption, all computer power is borne by the cloud server. In order to increase the scalability of the server, a Linux system is used as the underlying driver. Data translation Python 7 Data forward UDP The choice of cloud server is to balance the contradiction between large computing power and low power consumption. In order to increase the link duration, it is necessary to construct a virtual link by a large amount of computation. However, the low power consumption requirement makes the ground station unable to bear such a large computational burden. Remote computing by cloud server is the best solution

Access Link Assignments
To improve the efficiency of the ground station in the time resource and frequency resource, the time division multiple access (TDMA) strategy is preferred. In this strategy, we establish a schedule to manage connections between the satellites and multiple ground stations. As shown in Figure 10a, there are six steps in this strategy. Any data processed by the cloud data center will be stored for a long time. In the case of negligent operation, any transmitted data packets will be recorded in a log. For excellent performance, different software will be applied in our cloud data center, as shown in Table 3. For miniaturization and low power consumption, all computer power is borne by the cloud server. In order to increase the scalability of the server, a Linux system is used as the underlying driver. The choice of cloud server is to balance the contradiction between large computing power and low power consumption. In order to increase the link duration, it is necessary to construct a virtual link by a large amount of computation. However, the low power consumption requirement makes the ground station unable to bear such a large computational burden. Remote computing by cloud server is the best solution.

Access Link Assignments
To improve the efficiency of the ground station in the time resource and frequency resource, the time division multiple access (TDMA) strategy is preferred. In this strategy, we establish a schedule to manage connections between the satellites and multiple ground stations. As shown in Figure 10a, there are six steps in this strategy.

•
Access prediction: Calculate the connectable period between SSS-2A and each ground station with the LTE of SSS-2A. This calculation can be accomplished with the "ephem" library in python.

•
Overlap identification: As simulated in Section 4, many link durations may overlap each other. Identifying overlaps can increase assignment efficiency. • Construction of virtual link: For some special link durations, they can be combined into a virtual continuous link. • Schedule algorithm: Based on the above classification and construction, an efficient algorithm is the core of the entire strategy. • Driving mechanism: A lot of equipment in the ground stations needs to be warmed up. The driving mechanism can control the operation of the equipment according to the schedule. As shown in Table 4, the warm-up times of different devices are shown. Schedule algorithm: Based on the above classification and construction, an efficient algorithm is the core of the entire strategy.

•
Driving mechanism: A lot of equipment in the ground stations needs to be warmed up. The driving mechanism can control the operation of the equipment according to the schedule. As shown in Table 4, the warm-up times of different devices are shown. For the schedule algorithm, there are many algorithm structures available. Simplicity, stability and high efficiency are the most important. A Bellman-Ford algorithm is competent in this application because of high reliability. The pseudocode is shown in Figure 10b.
The implementation of virtual links relies on a high-performance schedule. According to the assignments, the ground station system can maximize the data transceiver capability. The establishment of virtual links is realized with the help of cloud computing.
Once the schedule has been established, the controller of the ground stations will run automatically. All received packets will be uploaded to the cloud data center. The TC packets will also be transmitted as required. These collaborative operations not only improve the link durations but also make adaptive communication possible.  For the schedule algorithm, there are many algorithm structures available. Simplicity, stability and high efficiency are the most important. A Bellman-Ford algorithm is competent in this application because of high reliability. The pseudocode is shown in Figure 10b.
The implementation of virtual links relies on a high-performance schedule. According to the assignments, the ground station system can maximize the data transceiver capability. The establishment of virtual links is realized with the help of cloud computing.
Once the schedule has been established, the controller of the ground stations will run automatically. All received packets will be uploaded to the cloud data center. The TC packets will also be transmitted as required. These collaborative operations not only improve the link durations but also make adaptive communication possible.

Adaptive Communication
An actual communication system has to compromise between performance and resource consumption. For microsatellites, the restricted power and frequency resource affect the throughput seriously. Dynamically adjusting the communication parameters can improve system efficiency. As shown in Figure 3, some important parameters are optional for synchronization, owing to the flexibility of the software-defined radio (SDR) structures. Different choices of those parameters make the system exhibit different performances. It can be seen from some contrasts that the 9600 bps data rate uses double the bandwidth than the 4800 bps data rate in the same situation. BPSK has better noise immunity than GMSK. However, the former has to contain equalizer and timing recovery modules. There are two different transmit intervals. We define the long interval as the slow packet mode and the short interval as the quick packet mode. The slow mode is more suitable than the quick mode in spare time. The choice of different coding methods is determined by the signal-to-noise ratio (SNR).
A connection is established with the initial state. The communication parameters will be adjusted by a self-loop between the satellite and ground station. As shown in Figure 11, when a satellite enters the coverage of the ground station, it will build a basic link as the initial mode. At the beginning, the most reliable communication parameters will be chosen. After the initial mode is stable, an adjust information TC packet and an adjust acknowledge TM packet are applied to accomplish the self-loop. Finally, the adjusted-state TM packets are transmitted according to the new communication parameters. In order to match the dynamic changes of the channel, such a self-loop will be repeated periodically during the entire connection period.
for synchronization, owing to the flexibility of the software-defined radio (SDR) structures. Different choices of those parameters make the system exhibit different performances. It can be seen from some contrasts that the 9600 bps data rate uses double the bandwidth than the 4800 bps data rate in the same situation. BPSK has better noise immunity than GMSK. However, the former has to contain equalizer and timing recovery modules. There are two different transmit intervals. We define the long interval as the slow packet mode and the short interval as the quick packet mode. The slow mode is more suitable than the quick mode in spare time. The choice of different coding methods is determined by the signal-to-noise ratio (SNR).
A connection is established with the initial state. The communication parameters will be adjusted by a self-loop between the satellite and ground station. As shown in Figure  11, when a satellite enters the coverage of the ground station, it will build a basic link as the initial mode. At the beginning, the most reliable communication parameters will be chosen. After the initial mode is stable, an adjust information TC packet and an adjust acknowledge TM packet are applied to accomplish the self-loop. Finally, the adjustedstate TM packets are transmitted according to the new communication parameters. In order to match the dynamic changes of the channel, such a self-loop will be repeated periodically during the entire connection period. Specific adaptive communications are accomplished by the interaction between the TM packets and TC packets. However, a complex decision algorithm is the core of an adaptive system. In this algorithm, the decision will be made according to distance, power resource, azimuth and pitch angle. The optimal solution is the key information of adaptation. For miniaturization and low power consumption, this algorithm will be embedded in the Aliyun Cloud server. Specific adaptive communications are accomplished by the interaction between the TM packets and TC packets. However, a complex decision algorithm is the core of an adaptive system. In this algorithm, the decision will be made according to distance, power resource, azimuth and pitch angle. The optimal solution is the key information of adaptation. For miniaturization and low power consumption, this algorithm will be embedded in the Aliyun Cloud server.
A satellite ground adaptive system can greatly improve communication efficiency. Adaptation must be based on the exchange of information. A feedback system between the satellite and ground station can improve the efficiency of the satellite ground communication system.

Design of Portable Ground Station
As described in Section 5, the networked ground station system consists of several independent ground stations. As shown in Figure 12, the unified design of the ground stations contains a digital frontend module, digital baseband module, radio frequency frontend module and power control module. Each module implements different functions, which work together to form a complete system.

Design of Portable Ground Station
As described in Section 5, the networked ground station system consists of several independent ground stations. As shown in Figure 12, the unified design of the ground stations contains a digital frontend module, digital baseband module, radio frequency frontend module and power control module. Each module implements different functions, which work together to form a complete system.

Digital Frontend Module
The conversion between digital signal and analog signal is an important process in the communication system. As shown in Figure 12, a digital frontend module is responsible for DAC/ADC, digital filter, digital resampling and another conversion function. The software-defined radio (SDR) is a radio communication system. The hardware of the SDR is replaced by an embedded system. For a portable design, the SDR is our best choice.
The USRP B210 SDR provides a fully integrated, single-board, universal software radio peripheral platform with continuous frequency coverage from 70 MHz to 6 GHz. The integrated RF frontend on the USRP B210 is designed with the new Analog Devices AD9361, a single-chip direct-conversion transceiver, capable of streaming up to 56 MHz of real-time RF bandwidth.
As shown in Figure 13, B210 contains two individual channels. It can satisfy both TC and TM functions simultaneously. A single B210 meets the digital conversion requirements of the entire ground station. However, B210 is only the hardware part of the SDR platform, and the driver software should be embedded in the digital baseband module.

Digital Frontend Module
The conversion between digital signal and analog signal is an important process in the communication system. As shown in Figure 12, a digital frontend module is responsible for DAC/ADC, digital filter, digital resampling and another conversion function. The softwaredefined radio (SDR) is a radio communication system. The hardware of the SDR is replaced by an embedded system. For a portable design, the SDR is our best choice.
The USRP B210 SDR provides a fully integrated, single-board, universal software radio peripheral platform with continuous frequency coverage from 70 MHz to 6 GHz. The integrated RF frontend on the USRP B210 is designed with the new Analog Devices AD9361, a single-chip direct-conversion transceiver, capable of streaming up to 56 MHz of real-time RF bandwidth.
As shown in Figure 13, B210 contains two individual channels. It can satisfy both TC and TM functions simultaneously. A single B210 meets the digital conversion requirements of the entire ground station. However, B210 is only the hardware part of the SDR platform, and the driver software should be embedded in the digital baseband module.
As described in Section 5, the networked ground station system consists of several independent ground stations. As shown in Figure 12, the unified design of the ground stations contains a digital frontend module, digital baseband module, radio frequency frontend module and power control module. Each module implements different functions, which work together to form a complete system.

Digital Frontend Module
The conversion between digital signal and analog signal is an important process in the communication system. As shown in Figure 12, a digital frontend module is responsible for DAC/ADC, digital filter, digital resampling and another conversion function. The software-defined radio (SDR) is a radio communication system. The hardware of the SDR is replaced by an embedded system. For a portable design, the SDR is our best choice.
The USRP B210 SDR provides a fully integrated, single-board, universal software radio peripheral platform with continuous frequency coverage from 70 MHz to 6 GHz. The integrated RF frontend on the USRP B210 is designed with the new Analog Devices AD9361, a single-chip direct-conversion transceiver, capable of streaming up to 56 MHz of real-time RF bandwidth.
As shown in Figure 13, B210 contains two individual channels. It can satisfy both TC and TM functions simultaneously. A single B210 meets the digital conversion requirements of the entire ground station. However, B210 is only the hardware part of the SDR platform, and the driver software should be embedded in the digital baseband module.

Digital Baseband Module
The core functions of a ground station are modulation/demodulation, coding/decoding, frequency capture, time correction and data forwarding. In SDR architectures, such core functions should be implemented in a microcomputer such as a PC, SoC and ARM. In our design, a Raspberry Pi 4 is chosen to undertake the digital baseband functions.
As shown in Table 5, there are six different parts in the digital baseband module. They undertake different functions. In particular, the demodulation blocks of BPSK and GMSK are specially designed by ourselves for better performance. The basic flows of demodulation and decoding are shown in Figure 14a, two different modulation types will be selected according to the control from the cloud computing script. Similarly, two different coding types will also be selected. The adaptive communication blocks are embedded in the demodulation modules.
As shown in Figure 14b, the B210 SDR platform and Raspberry Pi 4 are compact enough to be fixed in a mobile terminal. The power consumption is less than 5 W, which can be supplied by a power bank.

Digital Baseband Module
The core functions of a ground station are modulation/demodulation, coding/decoding, frequency capture, time correction and data forwarding. In SDR architectures, such core functions should be implemented in a microcomputer such as a PC, SoC and ARM. In our design, a Raspberry Pi 4 is chosen to undertake the digital baseband functions.
As shown in Table 5, there are six different parts in the digital baseband module. They undertake different functions. In particular, the demodulation blocks of BPSK and GMSK are specially designed by ourselves for better performance. The basic flows of demodulation and decoding are shown in Figure 14a, two different modulation types will be selected according to the control from the cloud computing script. Similarly, two different coding types will also be selected. The adaptive communication blocks are embedded in the demodulation modules.
As shown in Figure 14b, the B210 SDR platform and Raspberry Pi 4 are compact enough to be fixed in a mobile terminal. The power consumption is less than 5 W, which can be supplied by a power bank.

Radio Frequency Frontend Module
The radio frequency (RF) frontend is an important component in wireless communication systems. It contains antennas, filters, amplifiers, mixers and microwave networks. As shown in Figure 15, UHF is the downlink frequency, and VHF is the uplink frequency in our system. Each component in downlink and uplink plays a different role.

Radio Frequency Frontend Module
The radio frequency (RF) frontend is an important component in wireless communication systems. It contains antennas, filters, amplifiers, mixers and microwave networks. As shown in Figure 15, UHF is the downlink frequency, and VHF is the uplink frequency in our system. Each component in downlink and uplink plays a different role.  To compensate for the random attitude and low gain, a circularly polarized Yagi antenna is our best choice in the ground station. Each ground station contains a VHF circularly polarized Yagi antenna and a UHF circularly polarized Yagi antenna array. To track the satellites, a Yaesu G-5500 Azimuth-Elevation rotator system is applied. The G-5500 can make the antennas point precisely at the moving satellite. The cloud server calculates the pointing parameter through TLE and transmits it to the ground station for execution. To track our satellite continuously, multiple antennas should work together. As shown in Figure 16, several different antennas are distributed all over the country.

Antenna Design
A pair of linear polarized dipole antennas are mounted on the top of our microsatellite. To compensate for the random attitude and low gain, a circularly polarized Yagi antenna is our best choice in the ground station. Each ground station contains a VHF circularly polarized Yagi antenna and a UHF circularly polarized Yagi antenna array. To track the satellites, a Yaesu G-5500 Azimuth-Elevation rotator system is applied. The G-5500 can make the antennas point precisely at the moving satellite. The cloud server calculates the pointing parameter through TLE and transmits it to the ground station for execution. To track our satellite continuously, multiple antennas should work together. As shown in Figure 16, several different antennas are distributed all over the country.

Filter and Amplifier Design
The receive filter and transmit filter have different requirements in design. The former requires low insertion loss and high out-of-band rejection; the latter requires high power tolerance. Surface acoustic wave (SAW) devices are selected as the receive filter and cavity devices as the transmit filter.
As shown in Table 6, LNA requires a low-noise figure, and PA requires high output power. Because of the different voltages utilized by LNA and PA, a more efficient power control module is necessary.

Filter and Amplifier Design
The receive filter and transmit filter have different requirements in design. The former requires low insertion loss and high out-of-band rejection; the latter requires high power tolerance. Surface acoustic wave (SAW) devices are selected as the receive filter and cavity devices as the transmit filter.
As shown in Table 6, LNA requires a low-noise figure, and PA requires high output power. Because of the different voltages utilized by LNA and PA, a more efficient power control module is necessary.

Power Control Module
In order to increase energy efficiency, the power supply of the ground station should be controlled according to the cloud computing instructions. As shown in Figure 17a, when a satellite approaches the ground station, some equipment of the ground station will be powered gradually. Except for some networked equipment, other unrelated devices will be powered off during idle time. This strategy will greatly reduce unnecessary power consumption.

Power Control Module
In order to increase energy efficiency, the power supply of the ground station should be controlled according to the cloud computing instructions. As shown in Figure 17a, when a satellite approaches the ground station, some equipment of the ground station will be powered gradually. Except for some networked equipment, other unrelated devices will be powered off during idle time. This strategy will greatly reduce unnecessary power consumption. Power control commands are provided by cloud server and transmitted to Raspberry Pi 4 directly. As shown in Figure 17b, an Arduino microcontroller is used as the hardware of the power control module. Using the program developed by ourselves, various power control functions are possible in the ground station.
This module can greatly reduce the power consumption of the ground station. Complex processing is embedded in the cloud servers In this section, we introduced four modules, namely the digital frontend module, digital baseband module, RF front end module and power control module. The ground station is designed to meet the requirements of small phones and low power consumption.

Experiments and Verifications
To verify the credibility of the design, extensive experiments and data analysis are presented. During development, extensive simulation was used to aid the design. However, any simulations have uncertainty. On-orbit verification is the most convincing. In this paper, an on-orbit data service is implemented to verify the system.
The SSS-2A microsatellite was launched from the Taiyuan site at 19:05 Beijing time and successfully entered the predetermined orbit. According to calculations, the first entry into Chinese territory will be at 20:35 Beijing time, and a connection with the Xinjiang ground station will be established. The success of this connection determines the success of the entire mission. After the Xinjiang ground station, several connections can be established with Haerbin, Shanghai, Xi'an, Yunnan and Qinghai in tune. The performance of the networked ground stations can be verified in subsequent connections. Power control commands are provided by cloud server and transmitted to Raspberry Pi 4 directly. As shown in Figure 17b, an Arduino microcontroller is used as the hardware of the power control module. Using the program developed by ourselves, various power control functions are possible in the ground station.
This module can greatly reduce the power consumption of the ground station. Complex processing is embedded in the cloud servers.
In this section, we introduced four modules, namely the digital frontend module, digital baseband module, RF front end module and power control module. The ground station is designed to meet the requirements of small phones and low power consumption.

Experiments and Verifications
To verify the credibility of the design, extensive experiments and data analysis are presented. During development, extensive simulation was used to aid the design. However, any simulations have uncertainty. On-orbit verification is the most convincing. In this paper, an on-orbit data service is implemented to verify the system.
The SSS-2A microsatellite was launched from the Taiyuan site at 19:05 Beijing time and successfully entered the predetermined orbit. According to calculations, the first entry into Chinese territory will be at 20:35 Beijing time, and a connection with the Xinjiang ground station will be established. The success of this connection determines the success of the entire mission. After the Xinjiang ground station, several connections can be established with Haerbin, Shanghai, Xi'an, Yunnan and Qinghai in tune. The performance of the networked ground stations can be verified in subsequent connections.

Ground Station Experiments
Before SSS-2A enters the coverage of the ground station, all equipment will power on and warm up. According to our design, when the pitch angle is greater than 10 degrees, the ground station will be operated in the initial state. Once a connection is successfully established, the adaptive system starts.
As shown in Figure 18, the SDR platform receives the signal and displays it in the form of a spectrum. Figure 18a,b shows the received GMSK and BPSK signals, respectively. The former is the initial state of the system, and the latter is the result of adaptive adjustment. on and warm up. According to our design, when the pitch angle is greater than 10 degrees, the ground station will be operated in the initial state. Once a connection is successfully established, the adaptive system starts.
As shown in Figure 18, the SDR platform receives the signal and displays it in the form of a spectrum. Figure 18a,b shows the received GMSK and BPSK signals, respectively. The former is the initial state of the system, and the latter is the result of adaptive adjustment. As shown in Figure 19, the long and short intervals correspond to slow packets and quick packets, respectively. At the beginning, TM packets are transmitted in the initial state with a cycle of 8 s and switched to a cycle of 2 s according to the command. The switching of coding method and data rate will be reflected in subsequent data analysis. Once the signal demodulation is successful, the original data packets containing As shown in Figure 19, the long and short intervals correspond to slow packets and quick packets, respectively. At the beginning, TM packets are transmitted in the initial state with a cycle of 8 s and switched to a cycle of 2 s according to the command.

Ground Station Experiments
Before SSS-2A enters the coverage of the ground station, all equipment will power on and warm up. According to our design, when the pitch angle is greater than 10 degrees, the ground station will be operated in the initial state. Once a connection is successfully established, the adaptive system starts.
As shown in Figure 18, the SDR platform receives the signal and displays it in the form of a spectrum. Figure 18a,b shows the received GMSK and BPSK signals, respectively. The former is the initial state of the system, and the latter is the result of adaptive adjustment. As shown in Figure 19, the long and short intervals correspond to slow packets and quick packets, respectively. At the beginning, TM packets are transmitted in the initial state with a cycle of 8 s and switched to a cycle of 2 s according to the command. The switching of coding method and data rate will be reflected in subsequent data analysis. Once the signal demodulation is successful, the original data packets containing The switching of coding method and data rate will be reflected in subsequent data analysis. Once the signal demodulation is successful, the original data packets containing 256 bytes will be uploaded to the cloud server. It will be translated into numerical data in the cloud server immediately. Any authorized user can visit it through the UI interface.
As shown in Figure 20, the data packets of the attitude control system can fully indicate the orbit state and attitude state of SSS-A. All attitude control packets will be stored in the cloud server. For multiple satellites and multiple ground stations, the data packets from different ground nodes or satellites will be stored separately. The data of the integrated electronic system are shown in Figure 21a, which can also be visited on demand. As shown in Figure 21b, in order to clearly indicate the satellite statues, the data of the power system are given in the form of curves. Voltage, current and temperature are key parameters of the power system. The operational life of a satellite can be predicted by these parameters. Our ground station system satisfies the low power consumption requirements, and the total power consumption of the SDR and Raspberry Pi is less than 5 W. The average power consumption of the antenna system is also less than 5 W. Figure 21b, in order to clearly indicate the satellite statues, the data of the power system are given in the form of curves. Voltage, current and temperature are key parameters of the power system. The operational life of a satellite can be predicted by these parameters. Our ground station system satisfies the low power consumption requirements, and the total power consumption of the SDR and Raspberry Pi is less than 5 W. The average power consumption of the antenna system is also less than 5 W. As shown in Figure 21b, in order to clearly indicate the satellite statues, the data of the power system are given in the form of curves. Voltage, current and temperature are key parameters of the power system. The operational life of a satellite can be predicted by these parameters. Our ground station system satisfies the low power consumption requirements, and the total power consumption of the SDR and Raspberry Pi is less than 5 W. The average power consumption of the antenna system is also less than 5 W.

Networked Service Experiments
The reception performance can be increased by combining multiple ground nodes. As shown in Table 1, there are 18 opportunities for data reception in 24 h. Different numbers of data packets will be received at each link opportunity according to different link states.
As shown in Table 7, a data reception schedule can be constructed by the collaboration strategy. According to our analysis, link duration and pitch angle affect the reception capability severely. However, sometimes, these factors will compromise each other. Up to 2800 packets can be received in 24 h. This result is far beyond the data transmission capacity of the ordinary ground station. Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.