A proposal on an energy-saving policy on context based system

donde los servicios IPTV son definidos en perfiles basados en especificaciones como XSHTML, CSS and ECMASCRIPTS. [Matsubara, 2011] soporta un entono común interactivo que facilita el funcionamiento con servi cios IPTV pertenecientes a diferentes SPs. Sin embargo esta solución presenta una arquitectura con componentes rígidos los cuales limitan la posibilidad de conectar a proveedores con características especiales. Otros trabajos como [Nikola idis, 2006] presentan un modelo más flexible por el cual es posible hacer cambios en la configuración de dispositivos remotos de manera masiva acorde a unas características específicas. [Nikolaidis, 2006] y [Wey, 2009] propone usar TR-069 [BroadBand Forum, 2011] para transferir aplicacio nes y componentes software usando modelos y esquemas predefinidos que permiten generar la configuración apropia da para el dispositivo permitiendo reconfigurar parte del dispositivo. A pesar de ello no presenta un modelo de servi cio común en las capas superiores por lo que el modelo no es idóneo para poder trabajar con diferentes proveedores de servicios ya que cada proveedor de servicios no podría desa rrollar componentes software y puede que éstos no pudiesen ser ejecutados en el dispositivo por problemas de compatibi lidad. A pesar de las diferentes soluciones que tratan de resolver el problema de la interoperabilidad con diferentes SPs desde un mismo dispositivo, ninguna de ellas contempla una ar quitectura de control como se muestra en este paper. Existen trabajos que utilizan una arquitectura similar a la utilizada en este trabajo sobre diferentes escenarios. Por ejemplo, en [Zeller, 2009] se presenta un arquitectura de control muy similar a la que mostramos con el objetivo de controlar diferentes servicios software existentes en el automóvil. [Zeller, 2009] trata de controlar las ECUS (Electronic Con trol Units) del automóvil a través de los clústeres definidos y formando una jerárquica de capas de varios niveles. Para el trabajo presentado en este paper se ha optado por una arquitectura multicapa de tres niveles. Sin embargo, en la literatura se pueden observar diferentes soluciones con un número diferente de capas. Kramer y Magee [Kramer, 2007] también presenta una arquitectura genérica de tres capas entre las que se reparte las diferentes acciones a tener en cuenta (capa de gestión de cambios, capa de control y capa de gestión de objetivos). En cierta manera la configuración de Kramer y Magee tienen una cierta similitud con la distri bución presentada en este trabajo, sin embargo la decisión de tomada en la arquitectura que proponemos en este trabajo ha sido diseñada desde la distribución de las diferentes fun cionalidades identificadas para el dispositivo. 5 Implementación y Conclusiones Un prototipo básico ha sido implementado para testear el concepto de este trabajo. Los bloques fueron desarrollados en java y han sido instalados sobre un pequeño STB. El objetivo de este trabajo es proponer una arquitectura flexible para STB haciendo que sean capaces de adaptarse ellos mismos para satisfacer los requisitos técnicos de cada SP. Entre las ventajas que presentan cabe destacar la de reducir el número de dispositivos en los hogares de los usuarios en 18 el caso de que tengan contratados varios SP de IPTV. El sistema especifica un conjunto de servicios a tener en cuenta para facilitar el despliegue del sistema. Agradecimientos Los resultados de esta investigación han recibido financia ción del proyecto ARTEMISA TIN2009-14378-C02-02 y del proyecto HAUS IPT-2011-1049-430000 amparados dentro del plan nacional de J+D+i financiados por el Minis terio de Economía y Competitividad, Gobierno de España Reí erencias [BroadBand Forum, 2011] B. Forum, "TR-069 CPE W AN Management Protocol," 2011. [ITU, 2009], International Telecommunication Union, "JTU-T H.721", 2009. [Kramer, 2007], J. Kramer, J. Magee, "Self-Managed Sys tems: an Architectural Challenge", Future of Software Engineering, 2007. FOSE '07, vol., no., pp.259,268, 23- 25 May 2007. [Matsubara, 2011] F. M. Matsubara and M. Kawamori, "Lightweight interactive multimedia environment for TV," Consumer Electronics, IEEE Transactions on, vol. 57, pp. 283-287, 2011. [Nikolaidis, 2006] A. E. Nikolaidis, S. S. Papastefanos, G. l. Stassinopoulos, M. P. K. Drakos, and G. A. Doumenis, "Automating remote configuration mechanisms for home devices," Consumer Electronics, IEEE Transac tions on, vol. 52, pp. 407-413, 2006. [Wey, 2009]1. S. Wey, J. Luken, and J. Heiles, "Standardi zation Activities for IPTV Set-Top Box Remote Man agement," Internet Computing, IEEE, vol. 13, pp. 32-39, 2009. [Zeller, 2009], M. Zeller, G. Weiss, D. Eilers, R. Knorr, "A Multi-layered Control Architecture for Self-Management in Adaptive Automotive Systems," Adaptive and Intelli gent Systems, 2009. ICAIS '09. lnternational Conference on , vol., no., pp.63,68, 24-26 Sept. 2009. A proposal on an energy-saving policy on context based system Cuadrado Cordero, Ismael 1 Soria Morillo, Luis Miguei2 1 Scientific Computing Center of Andalusia (CICA) Economy, Innovation and Science Counseling Andalusia, Spain [email protected] 2Dept. of Languages and lnformatics Systems University of Seville Seville, Spain [email protected] Abstract the last decade, technology has created a new field 1 ware development, known as pervasive computing, nk lo the improvement and extension of integrated ti ors. These sensors have been successfully integrated lh daily-life of the common user, using devices like 1111 phones. Those new hardware improvements gave ommon developer the chance to generate new and 11 omplete applications, which interact with the user's wlth a mínimum effort. However, it also meant an 111 consumption of device's resources, which is 11 1 ted into a extra energy consumption. The 11 bution of this work is to propose a new policy for aving applications in the framework of context systems. The expected energy saving achieved 1111 h this police means to be significant both in the 1 tion of the duration of the battery by the user and 111 lng the impact of these devices in the environment. Keywords: Pervasive, eco, energy, saving l. Introduction 111 l devices experienced a boom after the appearance r I l generations' mobile phones and tablets. The ease of lopment and use and the wide range of applications 1 ble for those devices, as well as an affordable price, them a very attractive option for common users. Ion with the appearance and expansion of these devices, tl w business model based on mobile devices 1 1 otions was originated. Through buying and selling of 1 I otions, the user can create a device as personal as he 111 to, using only those applications that he considers 11111 rtant in his daily-life [5]. 1 /11 1/r la.1' XV Jornadas de ARCA, JARCA 2013 1 lt i/111.1· C'11alitativos y sus Aplicaciones en lf/,111110.v/.1·, Robótica e Inteligencia Ambiental / tlllr, /./. Muuros, J. A. Ortega, A. Fernández-Montes (eds.) I IIN1 V?N-84-6/6-7622-4 © 20/3 But the evolution of the hardware resources carried with it a stronger will from the developers of exploiting these resources and the users of having more accurate applications. However, these requests irnplied an extreme use of the rest of the device resources, meaning an exhausting use of energy [ 6] and, by hence, an important irnpact in the perception of the duration of the battery duration by the user and the C02 indirectly released to the atmosphere [7]. In order to adapt the hardware improvements to the real demands of the user, the development of new software should make an intelligent use of resources [8], besides the potential characteristics of the device. However, Due to the new model of business created for these environments, it is not possible to force a third party developer to make an accurate use of resources. Furthermore when the field of development directly involves the use of sensors and evaluation of multiple data [4], as it happens in the development of context-aware applications. Therefore, in this work we propose the use of a policy based system for a dynamic management of context aware applications in mobile devices [9]. Our approach uses a separated architecture of applications (context generator and user interface) and a DEA based system to classify the effectiveness of each one [ 1, 2]. 2. Our Approach The main idea of this work is to change the concept of development of Context-Aware applications from a closed developed application view to a collaborative system through developers. Splitting the user's interface to the background process of determining contexts, we can evaluate the effectiveness of the context's evaluation and choose the more adequate for our purposes. 19 - So, the first phase is to classify the contexts generators so they can be evaluated together. A context generator can provide for different contexts and it could be evaluated both as different single contexts and a more complex one. Once we defined the context, we have to define the input/output variables to be evaluated. This classification is done based on performance characteristics. Each context application performs one or more specific functions which can be classified according to the goal that each one of these functions is pursuing to achieve in order to solve a problem related with the user. According to these problems solved, two or more context can be classified between them as follows: •Similar contexts: If two or more context pursuits similar problems, for instance, two activity recognition services which addresses different activities (the first one determines the user's physical activity like walking, sitting and laying and the other one the user's virtual activity like working, driving or travelling), these context are addressed to be in the same category or 'similar'. •Equivalent contexts: In case that two or more similar contexts solve an equivalent problem but they are not exchangeable, for instance, two different activity recognition systems which solve a group of problems which are the same but include sorne other different activities (the first one determines the user's physical activity like walking, sitting and laying and the other one solves the same group of physical activities plus the user's virtual activity like working, driving or travelling), they are addressed to be 'intersection' or 'equivalent' contexts. •Exchangeable contexts: Finally, if two different systems provide a solution for the same group of problems, for instance, two different activity recognition systems which solve exactly the same activities. In this case, these contexts are addressed to be 'equal' or 'exchangeable' contexts. 3. Context-Aware Systems Pervasive computing technologies are based in offering 'anytime, anywhere, anyone' computing by decoupling users from devices To provide adequate service for the users, applications and services should be aware of their contexts and automatically adapt to their changing contexts-known as context-awareness. Context is any information that can be used to characterize the situation of an entity (like the present status of people, places, things and devices in the environment). An entity is a person, place, or object that is con idered relevant to the interaction between a user and an application, including location, time, activities, and th pr fi r n of u h nt.ity. 20 Context-aware systems are able to use contex information. A system is context-aware if it can extrae interpret and use context information and adapt i functionality to the current context of use. One goal of context-aware systems is to acquire an utilize information on the context of a device in order t provide services that are appropriate to the particul people, place, time, event, etc. Context-aware systems claim to be an important part o the future technologies' improvements in mobile an daily life systems. However, context-aware systems lac from a community of experienced developers in the systems and the excessive use of integrated sensors an hardware done by the existent systems and framewor affects the evolution ofthese systems and their integratio in devices with low hardware specifications. One of the main problems associated with contex aware systems is the energy use of different install applications in the device. Approaches to solve the issues are classified into hardware and software orient solutions. Hardware solutions are based on improving t use of energy made by hardware devices and they inclu bigger money budgets for development of solutio Software approaches include the better use of ener made by installed applications and the management running installed applications in the device by third p ones (like, for instance, the Operating System). 4. Data Envelopment Analysis (DEA) The methodology of DEA has been traditionally used estímate the relative efficiency of a set of producti units. In recent years, other applications have be developed in which includes DEA as a possible too! obtaining synthetic indices from partial indicators. It is originally a nonparametric procedure that uses linear programming technique that will allow evaluati of the relative efficiency of a set of homogeneo production units. The main advantage of this technique is targeted to flexibility, because it requires that all units grant the sa importance to the same partial indicator. Thanks to use of this too! for economical analysis we will be able evaluate the relative performance of the different ta and gives us a possible classification of suitability. For our work, Data Envelopment Analysis (DEA) been used as a software approach for reducing ene consumption. The use of DEA for our approach is done follows: A. Classification: Each context-aware system s be classified into a group and classified betw them. their equivalent and exchangeable contexts have been determined an evaluated between them a decision is taken: ' a. Equivalent contexts: In case that two or more equivalent contexts have been found in the system, it is necessary to determine ifthe information provided by both of them is being used in the system or one of them can be switched off b. Exchangeable contexts: In case there are two different context-aware systems installed in the device providing for the same solution, it is necessary to determine which one is suitable for the system_ (the less energy expense done, accordmg to the energy policy) and if the solution provided by them is used by the system or both can be switched off �. Conclusions 11111 xt-aware systems are in fashion nowadays and 1 111 to be improved due to the evolution of daily life 11 1 'devices and the possibility ofthe creation ofa new 1mmunity and the inclusion ofnew developers in it. 11 wever, the evolution of these systems should be 111 In time with an intelligent use of the resources and 1 t expenses. Our approach uses the Data 11 lopment Analysis (DEA) in order to determine the 1 1 V energy use of each application and manage them. , References An lntroduction to Efficiency and Productivity Is, Data Envelopment Analysis, Springer US, I , 78-0-387-24265-I, 10.1007/0-387-25895-7 6 1 1 hou, Peng, Beng Wah Ang, and Kim-Le� Poh. lll'V y of data envelopment analysis in energy and h lllmental studies. 11 European Journal of Operational 111 h 189.1 (2008): 1-18. 1 Tone, Kaoru. 11A slacks-based measure of 1 n y in data envelopment analysis. 11 European 11 1 of operational research 130.3 (2001): 498-509. 1 l l u, Moo-Ryong, et al. 11Energy-delay tradeoffs in tÍI hone applications. 11 Proceedings of the 8th rnutlonal conference on Mobile systems, applications, rvices. ACM, 2010. 1 1 1 aek, Jeongyeup, Joongheon Kim, and Ramesh lnd n. 11Energy-efficient rate-adaptive GPS-based 11 onlng for smartphones. 11 Proceedings of the 8th 1 111 llonal conference on Mobile systems, applications, 1 rvlces. ACM, 201 O. 1 1 Yoqub, Muhammad Farrukh, and lqbal Gondal. •11111 l phone based vehicle condition monitoring. 11 lu 11•1!1 lectronics and Applications (ICIEA), 2013 8th 11 onference on. IEE , 2013. B. Splitting: Each group is evaluated using D Envelopment Analysis and the relative use energy done by each context-aware system in group. For each one of these context-aw systems, a second evaluation is done according their equivalent and exchangeable contexts. O 111 Wong, Yi, et al. "A fram work of energy efficient 1 111 1 nsing for automat.i u r stot r cognition." lln ' of th 7th int rn 1tl n I onl' r n on Mobil m , nppli lllion , nnd rvl , A M, 2009. [8] Constandache, Jonut, et al. 11EnLoc: Energy efficient localization for mobile phones. 11 INFOCOM 2009, IEEE. IEEE, 2009. [9] Kang, Seungwoo, et al. "Seemon: scalable and energy-efficient context monitoring framework for sensor-rich mobile environments. 11 Proceedings of the 6th intemational conference on Mobile systems, applications, and services. ACM, 2008. 21