Search

Salvage surgery for recurrent oral squamous cell carcinoma (OSCC) often leads to a poor quality of life (QOL). The present study described three cases that resulted in favorable locoregional control with cetuximab in combination with radiotherapy (Cmab+RT). Case1 had regional recurrence of OSCC at the lower right mastoid area 4months after primary surgery. Case2 had regional recurrence of OSCC at the parotid area 7months after primary surgery. Case3 had local recurrence of OSCC at the masticatory muscle and Rouviere's lymph nodes 1year and 3months after primary surgery. In all cases, Cmab+RT was performed, and disease‑free survival was confirmed 4months, 2years and 6months, and 10months after Cmab+RT, respectively. Immunohistochemically, all resected tumors had no expression of 110‑kDa catalytic subunit of class IA phosphatidylinositol 3‑kinase (PI3Kp110α). In conclusion, if salvage surgery for recurrent OSCC results in a significantly low QOL, then shifting to chemoradiotherapy may be appropriate as a treatment strategy. In addition, strong evidence indicated that PI3Kp110α expression is associated with Cmab therapy efficacy., Oncology Letters, 24(6), art. no. 431; 2022

There is an urgent need to support high data rate among connected vehicles for both safety and infotainment purposes. However, high rate for vehicular communications is impacted by several challenges such as optimal beam tracking for multiple simultaneous-transmitting vehicles in highly-dynamic environments. In this paper, we aim to maximize the overall network throughput for multi-vehicular communications. We propose a reinforcement learning (RL) approach called combinatorial multi-armed bandit (CMAB) framework, where multiple arms (i.e., actions) form a super arm and can be played together, to handle the beam selection problem in a vehicular network. More specifically, we propose an adaptive combinatorial Thompson sampling algorithm, namely adaptive CTS, that CMAB embodies for appropriate selection of simultaneous beams in a high-mobility vehicular environment. The proposed approach applies the smart exploration-exploitation trade-off for the fast selection of beams. As the proposed adaptive CTS scheme produces a higher complexity over the search space which increases exponentially with the number of user, we also propose a sequential Thompson sampling (TS) algorithm where beam selection is performed in a one-by-one manner. We analyze the regret bounds for the proposed beam tracking algorithms. The performance of the proposed strategies are evaluated for multi-vehicle millimeter-wave (mmWave) simulation environments. Our results suggest that the proposed sequential approach performs almost similar to the simultaneous adaptive CTS scheme for tracking optimal beams in a multi-vehicular network with much reduced complexity. Simulation results also show that both of our proposed strategies approach the optimal achievable rate achieved by the genie-aided solution.

In this thesis, we study radio resource allocation (RRA) problems in fifth genera- tion (5G) systems with massive multiple-input multiple-output (MIMO) technol- ogy. We focus on optimizing the system performance (data rate maximization) of massive MIMO systems subject to quality of service (QoS) guarantees. However, these problems are extremely difficult to solve in massive MIMO, especially when practical challenges are taken into account, such as the need of a large number radio frequency (RF) chains, hybrid precoding and channel estima- tion. In order to solve the studied RRA problems in this thesis, we use as main tools optimization and contextual multi-armed bandits (CMAB). Also, this thesis is divided into two parts. The first part utilizes optimization to solve the problems of maximizing the data rate with and without considering QoS requirements. In this part, we propose a framework composed of three steps: clusterization, grouping, and scheduling. In the clusterization step, we create cluster of spatially compatible user equipments (UEs). In the grouping step, we select a set of space division multiple access (SDMA) groups from each cluster. In the scheduling step, we utilize these SDMA groups as candidates to receive resource blocks (RBs) aiming at solving a predefined RRA problem. We propose optimum and suboptimum solutions to solve the grouping and scheduling steps. The low-complexity proposed solutions present a good performance trade-off in relation to the highly complex optimal solutions and reference solutions. In the second part, we propose a framework utilizing dynamically adaptable CMAB to solve three RRA problems: i) data rate maximization; ii) data rate maximization with fairness guarantees, and; iii) data rate maximization with QoS guarantees, which are relevant problems in wireless communications. In this part, we utilize the clusterization and hybrid precoding to reduce the scheduling problem complexity by considering each cluster as an independent virtual CMAB scheduling agent. Next, we apply a new CMAB-based scheduler aiming to optimize the desired system performance metric. The solution for each problem utilizing our proposed framework is evaluated separately with UEs moving at different speeds. Simulation results showed that the proposed framework presents a good performance trade-off in data rate, fairness, and QoS in relation to the reference solutions. Nesta tese, estudamos os problemas de RRA (do inglês, radio resource allo- cation) em sistemas de quinta geração (5G) que utilizam a tecnologia MIMO (do inglês, multiple-input multiple-output) massivo. Focamos em resolver os pro- blemas de otimização de desempenho (maximização da taxa de dados) dos sistemas MIMO massivo sujeitos a garantias de QoS (do inglês, quality of ser- vice). Contudo, estes problemas são extremamente difíceis em MIMO massivo, principalmente quando levamos em consideração os desafios práticos, dos quais podemos citar a limitação no número de cadeias de Rádio-Frequência (RF), precodificação híbrida e estimação de canal. Nesta tese aplicamos otimiza- ção numérica e CMABs (CMABs, do inglês contextual multi-armed bandits) para solucionar problemas de RRA. Também, esta tese é dividida em duas partes. A primeira parte tem por objetivo utilizar otimização para resolver os problemas de maximização de taxa de dados considerando cenários com e sem requisitos de QoS. Nesta parte, propomos um arcabouço para solução do problema com- posto de três passos - clusterização, agrupamento, e escalonamento. No passo da clusterização, criamos clusters de UEs (do inglês, user equipment) espaci- almente compatíveis. No passo do agrupamento, selecionamos um conjunto de grupos SDMA (do inglês, space-division multiple access) de cada cluster. No passo do escalonamento, utilizamos estes grupos SDMA como candidatos para receber RBs (do inglês, resource blocks) com o objetivo de resolver um problema de RRA pré-definido. Nós propomos soluções ótimas e sub-ótimas para resolver os passos de agrupamento e escalonamento. As soluções propostas apresenta- ram um bom custo-benefício de desempenho em relação às soluções ótimas de alta complexidade e às soluções de referência. Na segunda parte, propomos um arcabouço utilizando CMAB dinamicamente adaptável para resolver três problemas de RRA: i) maximização da taxa de dados; ii) maximização da taxa de dados com garantias de justiça, e; iii) maximização da taxa de dados com garantias de QoS, que são problemas relevantes na área de comunicações sem fio. Nesta parte, utilizamos a clusterização e a precodificação híbrida para reduzir a complexidade do escalonamento considerando cada cluster como um agente de escalonamento CMAB virtual independente. Após isso, aplicamos um novo escalonador baseado em CMAB com objetivo de otimizar o desempenho desejado. Resultados de simulação mostram que o arcabouço proposto apre- senta um bom custo-benefício de desempenho em termos de taxa de dados, justiça e QoS em relação às soluções de referência.

We introduce a reinforcement learning algorithm inspired by the combinatorial multi-armed bandit problem to minimize the time-averaged energy cost at individual base stations (BSs), powered by various energy markets and local renewable energy sources, over a finite-time horizon. The algorithm sustains traffic demands by enabling sparse beamforming to schedule dynamic user-to-BS allocation and proactive energy provisioning at BSs to make ahead-of-time price-aware energy management decisions. Simulation results indicate a superior performance of the proposed algorithm in reducing the overall energy cost, as compared with recently proposed cooperative energy management designs.