Ȩ > ÀڷḶ´ç > ÀÚ·á°Ë»ö > Ç¥ÁØ
ÀÚ·á °Ë»ö°á°ú
Ç¥ÁØÁ¾·ù | Á¤º¸Åë½Å±â¼úº¸°í¼(TTAR) | ||||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Ç¥ÁعøÈ£ | TTAR-06.0273 | ±¸ Ç¥ÁعøÈ£ | |||||||||||||||
Á¦°³Á¤ÀÏ | 2023-10-19 | ÃÑ ÆäÀÌÁö | 28 | ||||||||||||||
ÇÑ±Û Ç¥Áظí | ¹«¼± ³×Æ®¿öÅ©ÀÇ ´Ù¸ñÀû ÀÚ¿ø°ü¸®¸¦ À§ÇÑ ºÐ»êÇü ÀΰøÁö´É Ȱ¿ë (±â¼úº¸°í¼) | ||||||||||||||||
¿µ¹® Ç¥Áظí | Decentralized Artificial Intelligence (AI) for Multi-Objective Resource Management in Wireless Networks (Technical Report) | ||||||||||||||||
ÇÑ±Û ³»¿ë¿ä¾à | ÀÌ ±â¼úº¸°í¼´Â ¹«¼± ³×Æ®¿öÅ© ÃÖÀûÈ À§ÇÑ ÀΰøÁö´ÉÀÇ »õ·Î¿î Ȱ¿ë »ç·Ê·Î ¹«¼± ³×Æ®¿öÅ©ÀÇ ´Ù¸ñÀû ÀÚ¿ø°ü¸® ÃÖÀûȸ¦ ¼öÇàÇÏ´Â ºÐ»êÇü ÀΰøÁö´É ¸ðµ¨ÀÇ ÈÆ·Ã ¹× Ãß·Ð ¹æ¹ýÀ» Á¦½ÃÇÑ´Ù. ¹«¼± ³×Æ®¿öÅ©ÀÇ ¸ðµç Á¢¼ÓÁ¡(access point: AP)µéÀÌ ¹«¼± ÀÚ¿øÀ» °øÀ¯Çϴ ȯ°æ¿¡¼, Á֯ļö È¿À²(spectral efficiency: SE) ¹× ¿¡³ÊÁö È¿À²(energy efficiency: EE)°ú °°Àº °¢ APÀÇ ¼·Î ´Ù¸¥ °³º° Çٽɼº´ÉÁöÇ¥ (key performance indicator: KPI)¸¦ µ¿½Ã¿¡ ÃÖÀûÈÇÏ´Â ´Ù¸ñÀû ÀÚ¿ø ÇÒ´ç ¹®Á¦¸¦ ½Ç½Ã°£À¸·Î µ¿ÀÛÇÏ´Â ºÐ»êÀûÀÎ ¿¬»êÀ¸·Î ÇØ°áÇÏ´Â ¹æ¹ýÀ» Á¦½ÃÇÑ´Ù. À̸¦ ÅëÇØ ´Ù¾çÇÑ Á¾·ùÀÇ Åë½Å ¼ºñ½º¸¦ ¿ä±¸ÇÏ´Â Â÷¼¼´ë ±â°è Åë½Å (machine-type communication: MTC) ½Ã½ºÅÛÀÇ ½Ç½Ã°£ ÀÚ¿ø°ü¸®¸¦ ¼öÇàÇÑ´Ù. ÀÌ ±â¼úº¸°í¼´Â 3GPP¿Í ORAN Ç¥ÁØ¿¡¼ Á¦½ÃµÈ AI/ML ±â¼úÀÇ ±Ô°ÝÀ» Âü°íÇÏ¿©, ºÐ»êÇü AI/ML ¸ðµ¨ÀÇ ÈÆ·Ã ¹æ¹ý°ú ¸ðµ¨ÀÇ ±¸ÇöÀ» À§ÇÑ ¹«¼± ÀåÄ¡°£ Çù·ÂÅë½Å ÀÎÅÍÆäÀ̽º °³³äÀ» ³íÀÇÇÑ´Ù. | ||||||||||||||||
¿µ¹® ³»¿ë¿ä¾à |
This technical report presents a method for training and reasoning of a distributed artificial intelligence model that performs multi-purpose resource management optimization as a promising use case of artificial intelligence for optimizing wireless networks. In an environment where all access points (APs) of a wireless network share radio resource, different individual key performance indicators of individual APs, such as spectral efficiency (SE) and energy efficiency (EE), can be addressed at the same time. A technical method is presented to solve the multi-purpose resource allocation problem that simultaneously optimizes performance indicators with distributed computation that operates in real time. Through this manner, real-time resource management of next-generation communication applications, in particular, machine-type communication systems requiring various types of communication services is conducted.
Based on the AI/ML technologies of 3GPP and ORAN standards, this technical report discusses the training method of the distributed AI/ML model and the concept of a cooperative communication interface between wireless devices for the implementation of the model. |
||||||||||||||||
°ü·Ã IPR È®¾à¼ | Á¢¼öµÈ IPR È®¾à¼ ¾øÀ½ | ||||||||||||||||
°ü·ÃÆÄÀÏ |
![]() |
||||||||||||||||
Ç¥ÁØÀÌ·Â |
|