A multichip module (MCM) packaging technology was developed for commercial and military applications based on the VLSI chips on silicon (VCOS) technology. This VCOS technology is extremely well suited for low-power, high-density CMOS applications where low cost and fast design cycles are critical. With the use of IBM's controlled collapse chip connection (C4), this technology can basically brick-wall good VLSI chips on the VCOS interconnecting substrate, thus achieving wafer-scale integration densities with 100% circuit yield. A special burn-in substrate can be used to burn-in and then test chips before mounting to the MCM, thus improving the final MCM assembly yield. This technology has been used in several military and commercial applications for both internal and external IBM programs. The VCOS technology is currently undergoing the certification and qualification of the Qualified Manufacturer Listing the US Department of Defense applications such as in the Advanced Spaceborne Computer Module program.<>
{"title":"Advanced silicon-on-silicon VLSI multichip packaging","authors":"D. Meyer","doi":"10.1109/NTC.1992.267905","DOIUrl":"https://doi.org/10.1109/NTC.1992.267905","url":null,"abstract":"A multichip module (MCM) packaging technology was developed for commercial and military applications based on the VLSI chips on silicon (VCOS) technology. This VCOS technology is extremely well suited for low-power, high-density CMOS applications where low cost and fast design cycles are critical. With the use of IBM's controlled collapse chip connection (C4), this technology can basically brick-wall good VLSI chips on the VCOS interconnecting substrate, thus achieving wafer-scale integration densities with 100% circuit yield. A special burn-in substrate can be used to burn-in and then test chips before mounting to the MCM, thus improving the final MCM assembly yield. This technology has been used in several military and commercial applications for both internal and external IBM programs. The VCOS technology is currently undergoing the certification and qualification of the Qualified Manufacturer Listing the US Department of Defense applications such as in the Advanced Spaceborne Computer Module program.<<ETX>>","PeriodicalId":448154,"journal":{"name":"[Proceedings] NTC-92: National Telesystems Conference","volume":"36 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1992-05-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115006486","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The Composite Analytic and Simulation Package for RFI (CASPR) is an analysis package that determines the performance of a coded signal in the presence of radio frequency interference (RFI) and additive white Gaussian noise. It can analyze a system with convolutional coding, Reed-Solomon coding, or a concatenation of the two. The signals can be either interleaved or noninterleaved. The model measures the system performance in terms of either the E/sub b//N/sub 0/ required to achieve a given bit error rate (BER) or the BER needed for a constant E/sub b//N/sub 0/.<>
{"title":"The Composite Analytic and Simulation Package for RFI (CASPR) on a coded channel","authors":"J. Freedman, T. Berman","doi":"10.1109/NTC.1992.267872","DOIUrl":"https://doi.org/10.1109/NTC.1992.267872","url":null,"abstract":"The Composite Analytic and Simulation Package for RFI (CASPR) is an analysis package that determines the performance of a coded signal in the presence of radio frequency interference (RFI) and additive white Gaussian noise. It can analyze a system with convolutional coding, Reed-Solomon coding, or a concatenation of the two. The signals can be either interleaved or noninterleaved. The model measures the system performance in terms of either the E/sub b//N/sub 0/ required to achieve a given bit error rate (BER) or the BER needed for a constant E/sub b//N/sub 0/.<<ETX>>","PeriodicalId":448154,"journal":{"name":"[Proceedings] NTC-92: National Telesystems Conference","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1992-05-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133889594","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The authors discuss the theory, implementation, and applications of the programmable digital receiver (PDRx) multichip module. The PDRx module comprises four digital receiver channels. The 9-in/sup 2/ module dissipates less than 25 W. The PDRx module accepts four 16-b-wide digital data streams at 33 MHz. The module then bandshifts, filters, and processes any signal in each input. These functions are all user programmable. The authors address the key innovative areas of the architecture as well as a number of applications. The applications include the digital demodulation of first, second, and M-level modulated signals, and cochannel interference reduction utilizing multiple receiver channels. The PDRx module contains multiple floating point processors which have a combined processing power of 100 million floating point operations per second. Utilizing these processors, the module supports a number of parallel processing configurations.<>
{"title":"Implementation of a programmable digital receiver multi-chip module","authors":"W.J. Rinard, D. Vujcic","doi":"10.1109/NTC.1992.267869","DOIUrl":"https://doi.org/10.1109/NTC.1992.267869","url":null,"abstract":"The authors discuss the theory, implementation, and applications of the programmable digital receiver (PDRx) multichip module. The PDRx module comprises four digital receiver channels. The 9-in/sup 2/ module dissipates less than 25 W. The PDRx module accepts four 16-b-wide digital data streams at 33 MHz. The module then bandshifts, filters, and processes any signal in each input. These functions are all user programmable. The authors address the key innovative areas of the architecture as well as a number of applications. The applications include the digital demodulation of first, second, and M-level modulated signals, and cochannel interference reduction utilizing multiple receiver channels. The PDRx module contains multiple floating point processors which have a combined processing power of 100 million floating point operations per second. Utilizing these processors, the module supports a number of parallel processing configurations.<<ETX>>","PeriodicalId":448154,"journal":{"name":"[Proceedings] NTC-92: National Telesystems Conference","volume":"146 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1992-05-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132293776","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
A multiphase program is in place to develop propagation experiments that will be conducted utilizing the beacons on the ACTS (Advanced Communications Technology Satellite). The beacons will transmit at 20 GHz and 27.5 GHz and will make it possible to acquire propagation data of great importance to the development of the next generation of communication satellites. The program began with experimentation with ESA's OLYMPUS satellite as a precursor to ACTS. Development of a prototype propagation terminal is nearing completion and construction of eight more terminals has begun. The process to select propagation experiments is underway. Successfully proposers will be supplied with NASA provided terminals to make in situ measurements. Experiments will also be chosen to carry out other types of investigations on topics such as the effects of the 20/30 GHz environment on mobile satellite applications. Mechanisms have been put in place to enlist the propagation research community in planning and coordinating the ACTS propagation experiments program.<>
{"title":"ACTS propagation experiments","authors":"J. Kiebler","doi":"10.1109/NTC.1992.267889","DOIUrl":"https://doi.org/10.1109/NTC.1992.267889","url":null,"abstract":"A multiphase program is in place to develop propagation experiments that will be conducted utilizing the beacons on the ACTS (Advanced Communications Technology Satellite). The beacons will transmit at 20 GHz and 27.5 GHz and will make it possible to acquire propagation data of great importance to the development of the next generation of communication satellites. The program began with experimentation with ESA's OLYMPUS satellite as a precursor to ACTS. Development of a prototype propagation terminal is nearing completion and construction of eight more terminals has begun. The process to select propagation experiments is underway. Successfully proposers will be supplied with NASA provided terminals to make in situ measurements. Experiments will also be chosen to carry out other types of investigations on topics such as the effects of the 20/30 GHz environment on mobile satellite applications. Mechanisms have been put in place to enlist the propagation research community in planning and coordinating the ACTS propagation experiments program.<<ETX>>","PeriodicalId":448154,"journal":{"name":"[Proceedings] NTC-92: National Telesystems Conference","volume":"35 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1992-05-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132525330","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The application of adaptive arrays to communications is addressed for the specific case of known pulsed desired signal transition times. From this information, the adaptive antenna weights that beam up on the desired signal can be determined. Previously proposed approaches work well when the thermal noise power level and interferer power levels are constant versus time, i.e., the scenario is stationary. But these approaches degrade seriously when the environment is not stationary. The authors present extensions to conventional processing to allow for near-ideal performance with large variations in thermal noise and/or interferer power levels. Computer simulations verify that the algorithm is effective even for very large (>20 dB) changes in power levels.<>
{"title":"Pulsed communication adaptive array processing","authors":"D. Farina, S. Kesler","doi":"10.1109/NTC.1992.267896","DOIUrl":"https://doi.org/10.1109/NTC.1992.267896","url":null,"abstract":"The application of adaptive arrays to communications is addressed for the specific case of known pulsed desired signal transition times. From this information, the adaptive antenna weights that beam up on the desired signal can be determined. Previously proposed approaches work well when the thermal noise power level and interferer power levels are constant versus time, i.e., the scenario is stationary. But these approaches degrade seriously when the environment is not stationary. The authors present extensions to conventional processing to allow for near-ideal performance with large variations in thermal noise and/or interferer power levels. Computer simulations verify that the algorithm is effective even for very large (>20 dB) changes in power levels.<<ETX>>","PeriodicalId":448154,"journal":{"name":"[Proceedings] NTC-92: National Telesystems Conference","volume":"70 2","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1992-05-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131809570","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Ongoing work by GEC-Marconi Electronic Systems Division in the development of a multifunction, common module, programmable digital radio/modem is described. The radio system handles conventional narrowband and spread-spectrum waveforms. The architecture developed is a programmable channelized approach. The modules of one channel are identical to those of the next, and as such are ideally suited to channel reusability during different parts of a mission or during failure modes of higher-priority functional systems. The hardware implementation exploits the latest RF, digital-RF, and digital signal processing component technologies. A PC-based brassboard was developed and used to demonstrate real-time programmable AM, FM, and IFF reception. The channelized architecture offers significant advantages over the fully integrated approach when only a few systems are to be integrated. Advantages are gained over the federated (a system at a time) approach when channel programmability/reusability is desired. The hardware implementation incorporating all digital modulation/demodulation techniques, coupled with the advent of software programmable devices, maintains the fully flexibility of the architecture.<>
{"title":"Programmable channelized digital radio","authors":"R. Prill, M. Antonesco","doi":"10.1109/NTC.1992.267861","DOIUrl":"https://doi.org/10.1109/NTC.1992.267861","url":null,"abstract":"Ongoing work by GEC-Marconi Electronic Systems Division in the development of a multifunction, common module, programmable digital radio/modem is described. The radio system handles conventional narrowband and spread-spectrum waveforms. The architecture developed is a programmable channelized approach. The modules of one channel are identical to those of the next, and as such are ideally suited to channel reusability during different parts of a mission or during failure modes of higher-priority functional systems. The hardware implementation exploits the latest RF, digital-RF, and digital signal processing component technologies. A PC-based brassboard was developed and used to demonstrate real-time programmable AM, FM, and IFF reception. The channelized architecture offers significant advantages over the fully integrated approach when only a few systems are to be integrated. Advantages are gained over the federated (a system at a time) approach when channel programmability/reusability is desired. The hardware implementation incorporating all digital modulation/demodulation techniques, coupled with the advent of software programmable devices, maintains the fully flexibility of the architecture.<<ETX>>","PeriodicalId":448154,"journal":{"name":"[Proceedings] NTC-92: National Telesystems Conference","volume":"18 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1992-05-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127778607","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
It is noted that the General Electric high density interconnect (GE-HDI) is an excellent solution to the digital multichip packaging problem, and it offers solutions to other difficult system interconnect problems in analog power, microwave, display, and sensors. GE-HDI offers the closest hybrid solution to wafer scale integration, achieving the benefits of performance and density without the limitations and yield problems of the monolithic approach. Relative to other hybrid applications, it offers significant advantages in performance and anticipated reliability due to its sputtered metal interconnection and excellent thermal properties while not sacrificing density or cost. Maskless adaptive lithography allows for rapid prototyping and flexibility in design and debugging. With the complete absence of tooling for mask or tape bonding, NRE costs are the lowest of any multichip module technology. HDI also requires the fewest processing steps, because there is no need to mount die on tape or to do lead or die bumping.<>
{"title":"GE high density interconnect: a solution to the system interconnect problem","authors":"M. Adler","doi":"10.1109/NTC.1992.267906","DOIUrl":"https://doi.org/10.1109/NTC.1992.267906","url":null,"abstract":"It is noted that the General Electric high density interconnect (GE-HDI) is an excellent solution to the digital multichip packaging problem, and it offers solutions to other difficult system interconnect problems in analog power, microwave, display, and sensors. GE-HDI offers the closest hybrid solution to wafer scale integration, achieving the benefits of performance and density without the limitations and yield problems of the monolithic approach. Relative to other hybrid applications, it offers significant advantages in performance and anticipated reliability due to its sputtered metal interconnection and excellent thermal properties while not sacrificing density or cost. Maskless adaptive lithography allows for rapid prototyping and flexibility in design and debugging. With the complete absence of tooling for mask or tape bonding, NRE costs are the lowest of any multichip module technology. HDI also requires the fewest processing steps, because there is no need to mount die on tape or to do lead or die bumping.<<ETX>>","PeriodicalId":448154,"journal":{"name":"[Proceedings] NTC-92: National Telesystems Conference","volume":"71 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1992-05-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126392798","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The general architecture and functionality of the MacTAC (Macintosh II based telemetry and command) system, including the individual telemetry modules, the various input-output interfaces, and the flexible user interface are described. This system performs telemetry data processing functions including frame synchronization, Reed-Solomon decoding, and packet reassembly at moderate data rates of up to 5 Mb/s sustained (20 Mb/s burst). The MacTAC is an easy-to-use, low-cost, transportable system designed to meet requirements specified by the Consultative Committee for Space Data Systems, while remaining flexible enough to support a wide variety of other space data processing requirements (e.g., time-division multiplexed data). In addition, the MacTAC can generate forward (uplink) data such as spacecraft commands, calculate and append a CRC, and output the data to NASCOM (NASA's ground communication network) to provide full telemetry and command capability. Semicustom VLSI gate arrays perform many of the return link (telemetry) functions of NASCOM deblocking, correlation, and frame synchronization. Reed-Solomon decoding and source packet reassembly are performed by modern microprocessor and semicustom VLSI components.<>
{"title":"Macintosh based space telemetry data acquisition system","authors":"W. K. Kay, C. T. Dominy, A. Collins","doi":"10.1109/NTC.1992.267874","DOIUrl":"https://doi.org/10.1109/NTC.1992.267874","url":null,"abstract":"The general architecture and functionality of the MacTAC (Macintosh II based telemetry and command) system, including the individual telemetry modules, the various input-output interfaces, and the flexible user interface are described. This system performs telemetry data processing functions including frame synchronization, Reed-Solomon decoding, and packet reassembly at moderate data rates of up to 5 Mb/s sustained (20 Mb/s burst). The MacTAC is an easy-to-use, low-cost, transportable system designed to meet requirements specified by the Consultative Committee for Space Data Systems, while remaining flexible enough to support a wide variety of other space data processing requirements (e.g., time-division multiplexed data). In addition, the MacTAC can generate forward (uplink) data such as spacecraft commands, calculate and append a CRC, and output the data to NASCOM (NASA's ground communication network) to provide full telemetry and command capability. Semicustom VLSI gate arrays perform many of the return link (telemetry) functions of NASCOM deblocking, correlation, and frame synchronization. Reed-Solomon decoding and source packet reassembly are performed by modern microprocessor and semicustom VLSI components.<<ETX>>","PeriodicalId":448154,"journal":{"name":"[Proceedings] NTC-92: National Telesystems Conference","volume":"85 12","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1992-05-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"120864777","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
A satellite and earth terminal systems study was initiated by the NASA Advanced Communication Technology Satellite (ACTS) Project Office and was performed by Hughes Network Systems. The purpose of the study was to investigate the potential of a specialized Ka-band spot beam satellite to reduce the cost of supervisory control and data acquisition (SCADA) communications. The study showed that the technology exists to build a system that could be priced less than $77 per month per RTU (remote terminal unit), including earth station equipment. The RTU earth station, or USAT, would cost $1500 at a quantity level of 500000. SCADA systems which poll RTU's rapidly will not be as economical to serve by satellite. A parallel study initiated by Southern California Edison and performed by Weber State University found similar satellite economics at Ku-band, but did not address earth station manufacturing cost. Market size estimates support a phased development approach. The first phase would be prototype system development at both Ku- and Ka-bands. NASA is currently sponsoring an additional study that would develop more detailed specifications.<>
{"title":"SCADA application for ACTS technology","authors":"B. Fairbanks","doi":"10.1109/NTC.1992.267919","DOIUrl":"https://doi.org/10.1109/NTC.1992.267919","url":null,"abstract":"A satellite and earth terminal systems study was initiated by the NASA Advanced Communication Technology Satellite (ACTS) Project Office and was performed by Hughes Network Systems. The purpose of the study was to investigate the potential of a specialized Ka-band spot beam satellite to reduce the cost of supervisory control and data acquisition (SCADA) communications. The study showed that the technology exists to build a system that could be priced less than $77 per month per RTU (remote terminal unit), including earth station equipment. The RTU earth station, or USAT, would cost $1500 at a quantity level of 500000. SCADA systems which poll RTU's rapidly will not be as economical to serve by satellite. A parallel study initiated by Southern California Edison and performed by Weber State University found similar satellite economics at Ku-band, but did not address earth station manufacturing cost. Market size estimates support a phased development approach. The first phase would be prototype system development at both Ku- and Ka-bands. NASA is currently sponsoring an additional study that would develop more detailed specifications.<<ETX>>","PeriodicalId":448154,"journal":{"name":"[Proceedings] NTC-92: National Telesystems Conference","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1992-05-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115869142","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The authors present radar detection, performance analysis results for targets in highly nonhomogeneous clutter environments where conventional adaptive array processing techniques are not feasible. For this case, the performance of an adaptive array processing technique which uses an angle-Doppler domain localized generalized likelihood ratio test is compared to that of a conventional displaced phase center aperture (DPCA) processor. When the measurement data available for estimation of clutter statistics is limited due to a severely nonhomogeneous environment, the conventional data-domain adaptive implementations of the joint domain optimal processor become undesirable because of their slow convergence (poor data efficiency) and heavy computation load. Under these conditions, the joint angle-Doppler domain localized generalized likelihood ratio test processor outperforms any conventional adaptive array processor, and is also shown to outperform the DPCA-based processor.<>
{"title":"A comparative study of clutter rejection techniques in airborne radar","authors":"M. Wicks, Hong Wang, L. Cai","doi":"10.1109/NTC.1992.267901","DOIUrl":"https://doi.org/10.1109/NTC.1992.267901","url":null,"abstract":"The authors present radar detection, performance analysis results for targets in highly nonhomogeneous clutter environments where conventional adaptive array processing techniques are not feasible. For this case, the performance of an adaptive array processing technique which uses an angle-Doppler domain localized generalized likelihood ratio test is compared to that of a conventional displaced phase center aperture (DPCA) processor. When the measurement data available for estimation of clutter statistics is limited due to a severely nonhomogeneous environment, the conventional data-domain adaptive implementations of the joint domain optimal processor become undesirable because of their slow convergence (poor data efficiency) and heavy computation load. Under these conditions, the joint angle-Doppler domain localized generalized likelihood ratio test processor outperforms any conventional adaptive array processor, and is also shown to outperform the DPCA-based processor.<<ETX>>","PeriodicalId":448154,"journal":{"name":"[Proceedings] NTC-92: National Telesystems Conference","volume":"52 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"1992-05-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129851273","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：