ÀÚ·á°Ë»ö-Ç¥ÁØ

Ȩ > ÀڷḶ´ç > ÀÚ·á°Ë»ö > Ç¥ÁØ

ÀÚ·á °Ë»ö°á°ú

°Ë»öÆäÀÌÁö·Î
Ç¥ÁØÁ¾·ù Á¤º¸Åë½Å±â¼úº¸°í¼­(TTAR)
Ç¥ÁعøÈ£ TTAR-06.0275 ±¸ Ç¥ÁعøÈ£
Á¦°³Á¤ÀÏ 2023-10-19 ÃÑ ÆäÀÌÁö 26
ÇÑ±Û Ç¥ÁØ¸í ¹«°æ°è ¼¿·ê·¯ ½Ã½ºÅÛÀÇ ¸ð¹ÙÀÏ ¿¡Áö ÄÄÇ»Æà ÀÚ¿ø ÇÒ´çÀ» À§ÇÑ ÀΰøÁö´É È°¿ë (±â¼úº¸°í¼­)
¿µ¹® Ç¥Áظí Artificial Intelligence (AI) for Resource Allocation in Cell-free Mobile Edge Computing System (Technical Report)
ÇÑ±Û ³»¿ë¿ä¾à ÀÌ ±â¼úº¸°í¼­´Â À̵¿Åë½Å ½Ã½ºÅÛÀÇ ¹«¼± ¾×¼¼½º¸¦ À§ÇÑ Àΰø Áö´É ¹× ±â°èÇнÀÀÇ »õ·Î¿î È°¿ë »ç·Ê·Î¼­, ÀΰøÁö´É ±â¼úÀ» ÀÌ¿ëÇÏ¿© ¹«°æ°è ¼¿·ê·¯ ³×Æ®¿öÅ©¿¡¼­ ¸ð¹ÙÀÏ ¿¡Áö ÄÄÇ»ÆÃ(mobile edge computing)À» À§ÇÑ Åë½Å ¹× ÄÄÇ»Æà ÀÚ¿øÀ» È¿À²ÀûÀ¸·Î Á¦¾îÇÏ´Â ¹æ½ÄÀ» Á¦½ÃÇÑ´Ù. ƯÈ÷, ¾ö°ÝÇÑ Áö¿¬ Á¦¾à Á¶°ÇÀ» ÃæÁ·Çϸ鼭 »ç¿ëÀÚÀÇ ¿¡³ÊÁö ¼Òºñ¸¦ ÃÖ¼ÒÈ­Çϱâ À§ÇÑ ¸ñÀûÀ¸·Î ´ÙÁßÀÇ ÀΰøÁö´É ¿¡ÀÌÀüÆ®µéÀÌ ºÐ»êÀûÀ¸·Î Çù·ÂÇÏ´Â ±¸Á¶¿Í À̸¦ À§ÇÑ ½ÉÃþ ÇнÀ ¾Ë°í¸®Áò°ú Á¦¾ÈÇÑ´Ù. À̶§ ºÐ»ê ¿¡ÀÌÀüÆ®°¡ ¼öÁýÇÏ´Â º¸»ó°ª(reward)À» Áß°£ ¼º´É ÁöÇ¥(intermediate KPI)·Î Çϸç, À̸¦ ÃÖ´ëÈ­ÇÏ´Â µ¿½Ã¿¡ ÃÖÁ¾ ¼º´ÉÁöÇ¥(eventual KPI)·Î¼­ ¿¡³ÊÁö È¿À²¼ºÀ» °í·ÁÇÑ´Ù. ÀÌ ±â¼úº¸°í¼­¿¡¼­ Á¦¾ÈÇÏ´Â ±¸Á¶¿Í ±¸Çö ¹æ½Ä¿¡ ÀÇÇØ °¢ »ç¿ëÀÚÀÇ °³º° Áö¿¬ Á¦¾à Á¶°ÇÀ» ¸¸Á·½ÃÅ°¸é¼­ ÀÌµé ¼º´É ÁöÇ¥ °ªµéÀ» ÃÖ´ëÈ­ ½Ãų ¼ö ÀÖ´Â °ÍÀ» È®ÀÎÇÑ´Ù. ÀÌ ±â¼úº¸°í¼­´Â 3GPP 5G NR ¹«¼±Á¢¼Ó ÀÎÅÍÆäÀ̽º¸¦ À§ÇÑ ÀΰøÁö´É/±â°èÇнÀ¿¡ ´ëÇÑ ±â¼ú º¸°í¼­ÀÎ 3GPP TR 38.843ÀÇ ÀϹÝÀûÀÎ ÇÁ·¹ÀÓ¿öÅ©¸¦ ÂüÁ¶ÇÏ¿©, ¹«¼±Á¢¼Ó ÀÎÅÍÆäÀ̽º¿Í ¿¬°èµÈ È°¿ë ¿¹½Ã(use case)¸¦ À§ÇØ ÇÊ¿äÇÑ ¼º´É Æò°¡ ¹æ¹ý·Ð°ú ÀýÂ÷¸¦ ±â¼úÇÑ´Ù.
¿µ¹® ³»¿ë¿ä¾à This technical report presents a novel application of AI/ML for wireless access in mobile communication systems. Specifically, it proposes a method to efficiently control communication and computing resources for mobile edge computing in a cell-free network using AI technology. The proposed approach involves a collaborative structure of multiple AI agents working in a distributed manner, along with deep learning algorithms, with the goal of minimizing users' energy consumption while meeting strict latency constraints for computing. The distributed agents collect reward values as intermediate key performance indicators (KPIs) and maximize them while considering energy efficiency as the eventual KPI. Through the proposed structure and implementation approach, it is observed that individual delay constraints of each user can be met while improving these performance indicator values. This report references the general framework of 3GPP TR 38.843, which is a technical report on artificial intelligence/machine learning for the NR air interface and provides performance evaluation methodologies and procedures necessary for its use cases.
°ü·Ã IPR È®¾à¼­ Á¢¼öµÈ IPR È®¾à¼­ ¾øÀ½
°ü·ÃÆÄÀÏ    TTAR-06.0275.pdf TTAR-06.0275.pdf
Ç¥ÁØÀÌ·Â
Ç¥Áظí Ç¥ÁعøÈ£ Á¦°³Á¤ÀÏ ±¸ºÐ À¯È¿
¿©ºÎ
IPR
È®¾à¼­
ÆÄÀÏ
¹«°æ°è ¼¿·ê·¯ ½Ã½ºÅÛÀÇ ¸ð¹ÙÀÏ ¿¡Áö ÄÄÇ»Æà ÀÚ¿ø ÇÒ´çÀ» À§ÇÑ ÀΰøÁö´É È°¿ë (±â¼úº¸°í¼­) TTAR-06.0275 2023-10-19 Á¦Á¤ À¯È¿ ¾øÀ½ TTAR-06.0275.pdf