M2M技術下的公眾分享設備之資源分配

Abstract

Resource allocation is a constant concern of engineering design and operations management. Because resources are always limited, it is important to design systems and processes which address resource assignment according to system’s capacity while satisfying the demand. When the consumer of such resources is a person and not a machine, consideration has to be given to the user’s perception of how long it takes to have a resource allocated. In this research, we address resource allocation problems in Publicly Shared Facilities with Deterministic service time (PSFD) where both efficient use of available resources and user perception of service availability have to be taken into account. These two factors are measured through the utilization ratio and the average wait time. In the Service Industry, organizations have implemented ways to allow the user to take an active role in the design of experience. This co-production of the user experience has shown that not only does the user perception of the service improves, but that in some cases the revenue also increases. Most of the time, user experience co-production has been enabled by those systems which give the user the knowledge of resource availability. Examples of those systems include laundry monitoring system LaundryView and resource scheduling system Disney MyMagic+. The latter, not only gives real time feedback of resource availability but it even allows the user to reserve the desired resources. PSFD could benefit from the features and standardization offered by M2M to enable user co-production systems, with the potential to improve utilization ratios and reduce user wait time. This research evaluated two M2M-enabled co-production systems, using a university dorm laundry room as a functional model. To study the effects on performance over standard operations by using M2M-enabled notification and reservation, three systems were modeled using Discrete Event Simulation (DES). A process-based DES was implemented using a purpose-specific Python module called SimPy. SimPy leverages on Python’s powerful generators feature to alternate execution among processes, a basic feature needed for process-based DES. All three models were compared based on average wait time, utilization ratio, and abandon ratio. Different scenarios were simulated to obtain the steady state values of three performance parameters. All three parameters were calculated for facilities with different number of resources and as a function of number of requests. Between the two M2M-enabled models, we conclude that the notification system had the greatest improvement on the performance parameters of a PSFD.

資源分配一直都是工程設計和營運管理考慮的重點。因為資源總是有限的，所以設計根據系統能力來分配資源，並同時滿足需求的系統和程序是很重要的。而當使用資源的是使用者而非機器時，考慮的則是在請求後經過多少時間使用者才能夠被分配到資源。

在這項研究中，我們專注於固定服務時間之公共設施(PSFD)的資源分配問題，其中必須同時考慮到資源之有效使用與使用者對服務可用度(Service Availability)之觀感。這兩項因子係透過使用率和平均等待時間來測量。在服務業中，業者已採用一些方法來讓使用者積極的參與使用經驗的創造。這種使用經驗的共同創造不僅能讓使用者對服務的滿意度提升，在某些情況下還能增加營收。大部分的時候，那些事先告知使用者資源可用度的系統都已導入了使用者經驗共同創造的功能。這類型系統的例子包含洗衣監測系統LaundryView和資源調度系統Disney MyMagic+。其中後者不只能即時反應資源的可用度甚至還允許使用者預約所需的資源。PSFD能從M2M標準化和其功能中受益，進而提供使用者經驗共同創造，並改善資源使用率和減少使用者的等待時間。

本研究以大學宿舍的洗衣間做為一個功能模型，評估兩個M2M的使用者經驗共同創造系統。分別透過研究M2M通知和預約功能來與在標準操作之下的系統作性能影響之比較。我們以離散事件模擬(DES) 技術來模擬此三系統之運作，我們使用流程式離散事件模擬(Process-based DES) 技術並採用一個叫作SimPy的特定用途Python模擬系統來實作。SimPy利用Python強大的產生器(generators) 功能在不同流程 (Process) 之間交替執行，這樣的交替執行是流程式離散事件模擬(Process-based DES) 的基本功能。三個系統以請求的平均等待時間、資源使用率以及放棄率來做為比較效能的依據。不同的情境都持續模擬直到取得穩定的三個效能參數。這三個效能數據都以不同數量的資源和請求來模擬以計算其數值. 在比較二個M2M模型後，我們的結論是，通知M2M系統對於PSFD的性能參數的改進有最大效果。