BEGIN:VCALENDAR VERSION:2.0 PRODID:IEEE vTools.Events//EN CALSCALE:GREGORIAN BEGIN:VTIMEZONE TZID:America/New_York BEGIN:DAYLIGHT DTSTART:20260308T030000 TZOFFSETFROM:-0500 TZOFFSETTO:-0400 RRULE:FREQ=YEARLY;BYDAY=2SU;BYMONTH=3 TZNAME:EDT END:DAYLIGHT BEGIN:STANDARD DTSTART:20251102T010000 TZOFFSETFROM:-0400 TZOFFSETTO:-0500 RRULE:FREQ=YEARLY;BYDAY=1SU;BYMONTH=11 TZNAME:EST END:STANDARD END:VTIMEZONE BEGIN:VEVENT DTSTAMP:20260220T144647Z UID:7EAC5609-9B05-4575-8E8B-C5F6BB550445 DTSTART;TZID=America/New_York:20260219T180000 DTEND;TZID=America/New_York:20260219T190100 DESCRIPTION:Emerging latency-sensitive applications such as autonomous driv ing\, immersive extended reality\, and large-scale sensing are expected to impose more stringent and heterogeneous requirements on beyond-5G/6G comm unication systems. These applications simultaneously demand ultra-low\nlat ency\, high reliability\, and task-specific semantic fidelity under tight bandwidth and computational constraints\, which are challenging to address with conventional bit level communication paradigms. Additionally\, as cu rrent wireless communication systems approach the Shannon capacity limit\, coupled with the explosive growth of connected devices with data-intensiv e applications\, making further improvements in transmission efficiency is becoming increasingly critical for future wireless networks. In recent ye ars\, semantic communication (SemCom) has emerged as a promising paradigm to bridge this gap by transmitting semantic information (SI) that is most relevant to the downstream task or communication goal and discarding task- irrelevant information to improve transmission efficiency and reduce commu nication overhead. However\, due to the complex task objectives and dynami c environments\, the unification of SemCom into current wireless networks is plagued by a plethora of challenges\, including but not limited to meet ing the diverse needs of heterogeneous users in dynamic conditions and int elligent resource allocation to ensure efficient utilization especially in resource constrained scenarios. Accordingly\, this thesis develops novel SemCom frameworks to address these challenges.\nFirstly\, to address the c hallenges posed by diverse channel conditions\, changing user task perform ance demands\, and limited computational capabilities amongst users\, this thesis proposes CAAR-MTSC (Context-Aware\, Adaptive-rate\, Multi-Task Sem antic Communication)\, a\ndeep learning-enabled SemCom architecture that s upports concurrent downstream tasks. Central to the design is a Semantic M ask Module (SMM) with its two-fold functionality to preserve task-salient SI and discard redundant SI (semantic compression) depending on the dynami c channel conditions and the user’s perceptual need. We then introduce a novel composite rate–distortion– perception (RDP) objective that find s the tradeoff between perceptual quality and data reconstruction. Then\, a knowledge distillation (KD) paradigm is introduced to train a lightweigh t receiver for resource-constrained devices\, while relatively retaining t he high downstream task performance of the larger computationally intensiv e receiver. Simulation results prove that the proposed framework is able t o achieve high concurrent multi-task accuracy under dynamic channel condit ions with changing user needs on resource-constrained devices.\nSecondly\, to address the combinatorial challenges faced with heterogeneous users wi th diverse multi-tasking and latency needs in a resource constrained wirel ess network\, we introduce a multi-task multi-user semantic communication system to allocate wireless resources and select appropriate degree of sem antic compression to meet the needs of heterogeneous users. A novel Segmen tation-Guided Dual-Branch Encoder for Multi-Task Semantic Communication (S egDB-MTSC) architecture is proposed to individually extract SI from region s of interest (ROI) and regions of non-interest (RONI). A latency-aware mu lti-task semantic utility (LAMTSU) term is derived to maximize the compreh ensive system performance for diverse users\, including their multi-task p erformance and their latency constraints. A deep reinforcement learning (D RL) algorithm is then used to perform resource allocation to maximize long term system utility. Simulation results illustrate the superiority of the proposed scheme to meet the latency and task needs of users.\n\nVirtual: https://events.vtools.ieee.org/m/540170 LOCATION:Virtual: https://events.vtools.ieee.org/m/540170 ORGANIZER:fchen324@uwo.ca SEQUENCE:35 SUMMARY:Goal-Oriented Multi-Task Semantic Communication for Efficient Wirel ess Transmission under Dynamic Channels in a Resource Constrained Environm ent URL;VALUE=URI:https://events.vtools.ieee.org/m/540170 X-ALT-DESC:Description: <br /><div dir="auto" data-olk-copy-source="Message Body">Emerging latency-sensitive applications such as autonomous driving\, immersive extended reality\,&nbsp\;and large-scale sensing are expected t o impose more stringent and heterogeneous requirements&nbsp\;on beyond-5G/ 6G communication systems. These applications simultaneously demand ultra-l ow</div>\n<div dir="auto">latency\, high reliability\, and task-specific s emantic fidelity under tight bandwidth and&nbsp\;computational constraints \, which are challenging to address with conventional bit level&nbsp\;comm unication paradigms. Additionally\, as current wireless communication syst ems approach the&nbsp\;Shannon capacity limit\, coupled with the explosive growth of connected devices with data-intensive applications\, making fur ther improvements in transmission efficiency is becoming&nbsp\;increasingl y critical for future wireless networks. In recent years\, semantic commun ication&nbsp\;(SemCom) has emerged as a promising paradigm to bridge this gap by transmitting semantic&nbsp\;information (SI) that is most relevant to the downstream task or communication goal and&nbsp\;discarding task-irr elevant information to improve transmission efficiency and reduce&nbsp\;co mmunication overhead. However\, due to the complex task objectives and dyn amic&nbsp\;environments\, the unification of SemCom into current wireless networks is plagued by a plethora of&nbsp\;challenges\, including but not limited to meeting the diverse needs of heterogeneous users in&nbsp\;dynam ic conditions and intelligent resource allocation to ensure efficient util ization especially in&nbsp\;resource constrained scenarios. Accordingly\, this thesis develops novel SemCom frameworks to&nbsp\;address these challe nges.</div>\n<div dir="auto">Firstly\, to address the challenges posed by diverse channel conditions\, changing user task&nbsp\;performance demands\ , and limited computational capabilities amongst users\, this thesis&nbsp\ ;proposes CAAR-MTSC (Context-Aware\, Adaptive-rate\, Multi-Task Semantic C ommunication)\, a</div>\n<div dir="auto">deep learning-enabled SemCom arch itecture that supports concurrent downstream tasks. Central&nbsp\;to the d esign is a Semantic Mask Module (SMM) with its two-fold functionality to p reserve task-salient SI and discard redundant SI (semantic compression) de pending on the dynamic channel&nbsp\;conditions and the user&rsquo\;s perc eptual need. We then introduce a novel composite rate&ndash\;distortion&nd ash\;&nbsp\;perception (RDP) objective that finds the tradeoff between per ceptual quality and data&nbsp\;reconstruction. Then\, a knowledge distilla tion (KD) paradigm is introduced to train a lightweight&nbsp\;receiver for resource-constrained devices\, while relatively retaining the high downst ream task&nbsp\;performance of the larger computationally intensive receiv er. Simulation results prove that the&nbsp\;proposed framework is able to achieve high concurrent multi-task accuracy under dynamic&nbsp\;channel co nditions with changing user needs on resource-constrained devices.</div>\n <div dir="auto">Secondly\, to address the combinatorial challenges faced w ith heterogeneous users with diverse&nbsp\;multi-tasking and latency needs in a resource constrained wireless network\, we introduce a multi-task mu lti-user semantic communication system to allocate wireless resources and select&nbsp\;appropriate degree of semantic compression to meet the needs of heterogeneous users. A novel&nbsp\;Segmentation-Guided Dual-Branch Enco der for Multi-Task Semantic Communication (SegDB-MTSC) architecture is pro posed to individually extract SI from regions of interest (ROI) and region s&nbsp\;of non-interest (RONI). A latency-aware multi-task semantic utilit y (LAMTSU) term is derived to&nbsp\;maximize the comprehensive system perf ormance for diverse users\, including their multi-task&nbsp\;performance a nd their latency constraints. A deep reinforcement learning (DRL) algorith m is then used to perform resource allocation to maximize long term system utility. Simulation results&nbsp\;illustrate the superiority of the propo sed scheme to meet the latency and task needs of users.</div> END:VEVENT END:VCALENDAR