Page 146 - ITU KALEIDOSCOPE, ATLANTA 2019

2019 ITU Kaleidoscope Academic Conference

• Self-repairing artificial fish swarm algorithm: Owing to multi-objective knapsack problem, a global optimization
the stronger optimization ability and faster convergence algorithm is urgently needed to search for an optimized SDSP,
speed, we proposed the self-repairing artificial fish which promotes the accuracy of smart services and reduces
swarm algorithm(SAFSA) to search for an optimized the cost of sensing devices. For a multi-objective knapsack
SDSP by limiting artificial fish moving near the problem, people have put forward valuable methods, mainly
constraint boundary by self-repairing behavior. divided into two types: (1) heuristic algorithms, such
as greedy algorithm, dynamic programming algorithm,
The rest of this paper is organized as follows: we briefly simulated annealing algorithm and so on; (2) swarm
review the past research on the in-home health-care system intelligent optimization algorithm: genetic algorithm [17],
in the smart home, and intelligent optimization algorithms particle swarm optimization algorithm [18], ant colony
in section 2. We propose the framework of the APGS in algorithm [19] and so on. Because of the low efficiency
section 3. The Smart-desire mapping method is described in and slow convergence rate for large-scale problems, heuristic
section 4 and self-repairing artificial fish swarm algorithm is algorithms are replaced by swarm intelligent optimization
presented in section 5. We perform a series of experiments algorithm. However, classical intelligent optimization
to verify the scientificity and validity of SDSP in section 6. algorithms will usually plunge into local optimization and
The conclusion and future work are discussed in section 7. are sensitive to the initial parameters. For example, the
artificial fish swarm may plunge into local optimization if
2. RELATED WORKS the visual of artificial fish is too small; additionally, changes
of crossover and mutation probabilities of genetic algorithm
Recently, owing to the development and advancement of the will result in different genetic speeds. Therefore, in order to
Internet of things and digital health, research in the smart generate the optimized SDSP for geriatric care, we present a
home becomes increasingly important. A smart home, in self-repairing artificial fish swarm algorithm, which introduce
which artificial intelligence techniques control home settings, self-repairing behavior to limit artificial fish searching for an
collects data by sensors when residents perform their normal optimized solution near the constraint boundary.
daily routines. Since sensors can collect data in a naturalistic
way without modifying an individual’s behavior, the smart
home provides a new way for automated health care. The 3. THE FRAMEWORK OF AUTOMATIC PLAN
survey in [10] showed that all participants had positive GENERATING SYSTEM
attitudes towards the technology of the smart home and were
willing to accept the installation of sensing devices in their As shown in Figure 1, we formalize user care demands
homes. Afterwards, more researchers were committed to and expert knowledge into digital description based on
providing health care services for users of various ages. Portet expert diagnosis, medical literature, and clinical diagnosis.
et al. [11] showed that inexpensive smart home technologies Additionally, smart services for geriatric care are extracted
could be used for the purpose of self-monitoring of safety, from recent researches. In this framework, Automatic Plan
health and functional statuses in existing homes, and are Generating is the key module including two submodules: (1)
urgently required. Nehmer et al. [12] used the smart home Smart-desire module is proposed to extract required smart
to provide a better assistant system in health monitoring and services for geriatric care by decomposing care demands
to improve the quality of life of elderly and disabled people. into atomic demands, calculating functional similarity and
Skubic et al. [13] provided passive sensor networks to capture non-functional similarity between atomic demands and smart
patterns representing physical and cognitive health conditions service; (2) Global optimization module, in which the SAFSA
in an aging in place elderly-care facility. Additionally, Mario is proposed to search for the optimized solution of SDSP
et al. [14] proposed a software architecture that modeled the for geriatric care based on cost evaluation and performance
functionalities of a smart home platform to deploy sensitive evaluation.
services into the digital home for health care.
However, due to the high costs and untrusted design, the Sensing Device Decoding Optimized
smart home failed to make headway in the field of health Elderly User Selection Plan Solution
care. In order to reduce the costs and to systematize the Care Demands Decomposition
Traveling Expert Knowledge Forest
design of SDSP, we first formalize elderly care demands and Judging Slot Values Smart-Desire Global Optimization
Algorithm
smart services in a fixed data structure based on medical Expert Diagnosis, Medical Literature and Clinical Diagnosis User Demands Extracting Atomic care demands Self-repairing Artificial Global Optimization
Fish Swarm Algorithm
diagnosis and recent research in the smart home, such as
Problem
activity recognition, health change detection, falling detection Functional Non-functional Selected services Conversion
Similarity
Similarity
and so on [15]. Since the expert knowledge system with a Expert Knowledge Unknown Data
Acquisition
large amount of knowledge provided a method of simulating Recent Researches TF-IDF Statistics Semantic Similarity Calculation Intuitionistic Fuzzy Cosine similarity Knapsack Probem
Multi-objective
the decision process of human experts [16], the expert Smart Services
knowledge learned in geriatric diagnosis was adopted to Automatic Plan Generating
decompose user demands into atomic demands. Thereafter, Figure 1 – The framework of APGS
we extracted smart services based on functional similarity and
QoS similarity between atomic demands and smart services. More narrowly, user demands are decomposed into atomic
As the selection of sensing devices is regarded as a care demands automatically by traveling expert knowledge

– 126 –