Hybrid Aspect of Context-Aware Middleware for Pervasive Smart Environment: A Review

Pervasive computing hasmade almost every device we see today to be communicated and function in collaboration with one another. Since the portable devices have become a part of our everyday life, people are more involved in a pervasive computing environment. *ey engage with many computational devices simultaneously without knowing the availability of their existence.*e current world is being filled with more and more smart environments. *ese smart environments make them to be attracted towards the new technological emergence in the field of pervasive computing. Various researches are being carried out to improve the smart environment and their applications. Middleware plays a vital role in building the pervasive applications.*e pervasive devices act based on the context of the situation, that is, they do their actions according to the environment of the application. *ey react to the situations smartly as they can take their own decisions based on the context developed for that specific application. Most of the pervasive applications were using its ownmiddleware that is specific towards their need. As today, most of the applications are using their own middleware with their specific requirement, which leads us to unearth out their common features and their scope of using it. In this paper, a survey on the various hybrid aspects of the different context-aware middleware has been done.*is middleware is classified based on service, context, and device aspects. Merits and demerits are identified from the existing smart environments, and future perspective of their development such as generic context-aware middleware need has been discussed.


Introduction
Pervasive computing is drastically shifting the day-to-day activities by bringing "anytime, anywhere, and anything" computing potential into the living environment.Computers spread across different environments and tend to merge and disappear into day-to-day objects [1].ey are capable of sensing the context of the environment, communicating with different objects and providing the right information to the user.Due to these rich computing powers, more pervasive computing applications are developed, and the need of the pervasive application is sensed today.As the demand of pervasive application grows enormously, more and more research are undertaken in the industry and in academia.
Most of the researches in pervasive computing focus on addressing the specific domain application development, for example, smart home-like aware home [2], UbiHome [3], and inkHome [4].e objectives of these studies are to pattern the user behavior and to provide different services for matching the pattern that supports the "comfort of the user."From the existing applications, researchers identified the major challenge like "how to handle different devices and different contexts" and also proposed middleware as a solution to it.

Evolutions of Pervasive Computing
According to Gabriel et al. [5], before three decades, "computer" is totally referred as mainframe systems which simply processes input and produces output.is period is referred as the rst wave of computing where the user and computer will have less interaction.e rst era of computing is termed as "many users-one computer"; that is, one computer is shared by many users.By the rapid growth in the computing vicinity, personal computer came into existence.is period is referred as the second wave of computing, and it is termed as "one user-one computer"; that is, one user uses one computer.Personal computer is very powerful as a standalone system, and it takes part as a node when the network is evolved.
Enormous growth in smart devices/gadgets gave rise to the new computing dreams called ubiquitous computing; this was initiated by Mark Weiser who is called as the father of ubiquitous computing.is period is referred as the third wave of computing, and it is also termed as "one user-many devices"; that is, one user is supported by many devices such as computer, smart phone, and tab as shown in Figure 1.
Ubiquitous computing is a combination of pervasive computing and mobile computing [6].Making the computing service available in any location context is the primary objective of mobile computing, whereas the primary objective of pervasive computing is to acquire the context information from the environment.
Ultimate focus of mobile computing and pervasive computing is on context-aware computing [7,8].So, the term "ubiquitous computing" is also called as pervasive computing, and it can be used either way [9,10].Ubiquitous computing is also referred as disappearing computing or everywhere computing or ambient intelligence or invisible computing or internet of things [11].
Pervasive computing has slowly and steadily taken the computing world into a new dimension with its features such as adaptive system, location sensitivity, energy-aware system, and context-aware system.

Pervasive Applications
ere is a rapid growth in the device world where most of the devices support smartness and the environments are constructed with the intention of supporting smartness.By the integration of smart devices (e.g., microdevices and complex devices) and smart environments (e.g., embedded devices in the physical environment such that a wall can sense camera is recording), a new environment is created that supports any environment and any device, which will work together to produce the pervasive smart environment as shown in Figure 2.

Classifications of Pervasive Architecture
Architecture is the rst step in the development of any system; pervasive architecture is classi ed into three types based on its function, data, and its working as shown in Figure 3 [8].

Direct Sensor Access Architecture.
is architecture was widely used in the earlier stage of pervasive computing when devices were available for direct access.e client software gathers the required information from the available devices without any additional layer to acquire the information from the device.Separate drivers are hardcoded into the speci c application.Hence, this can be used in di erent applications as there will be di erent drivers for di erent devices.
erefore, this architecture is well suited for standalone systems which will directly access the devices as it is capable of handling homogeneous devices.More studies on pervasive architecture are being conducted in different dimensions such as "interoperability of service," "interoperability of the device," "context modeling," and "building smart environment".e primary objective of pervasive computing is "context-aware computing" which deals with the different contexts of the application like location context, user context, and time context.ese contexts are helpful to provide appropriate service to the user.Context can be defined as "any information used to predict the state of an object, where the object can be user, location, or any physical object."Today, middleware for context-aware computing is one of the further steps for pervasive computing to move towards veracity.

Related Works on Context-Aware Middleware
According to Bandyopadhyay et al. [21], context-aware middleware is essential for three reasons: (i) To develop a common method to interact between heterogeneous devices belonging to a different domain.(ii) Middleware acts as a relationship to merge different heterogeneous components together.(iii) Hiding of complexity can be achieved in different applications with different domains.
Last two decades have been the peak periods in the development of context-aware middleware [22], where most of the middleware are classified into seven categories [23]: agentbased middleware, reflective middleware, metadata-based middleware, tuple space middleware, adaptive middleware, objective middleware, and OSGI-based middleware.
ere are several surveys carried out in relation to the context-aware middleware field.e study and analysis of the survey are given in Table 1.Table 1 specifies the functional and nonfunctional characteristics of the existing context-aware middleware based on literature study [7,24].
As per the result of the survey mentioned in Table 1, the working aspects of middleware can be classified into three levels of aspects, namely, service aspect, context aspect, and device aspect, from the context perspective as shown in Figure 4. Service aspects act as an interface between the application layer and the context aspects.It converts the requirement into service.Context aspects play a vital role in bridging the gap between the service aspects and the device aspects, where context information is derived from the device aspects.Device aspects act as a logical device to share information between context aspects and physical devices.
From the literature survey, it is obvious that the contextaware middleware is growing as an indispensable architecture component in building any pervasive application.From the study mentioned in Table 1, important parameters used for evaluating any context-aware middleware are identified as device management, interoperability, context awareness, security and privacy, autonomy, intelligence, and adaptability.
ese parameters are used to analyze the context-aware middleware.
A study on the various context-aware middleware is carried out, in order to understand the working of different aspects of the context-aware middleware and to make sure the number of domains that the context-aware middleware supports.Metrics for evaluating the existing context-aware middleware are achieved through the number of domains that the same middleware can support.e "multi" effects (multidomain, multiapplication, and multiculture) are to be addressed at the architecture level [32,34].In order to categorize the existing context-aware middleware, the following criteria are proposed: (i) If it supports three or more domains, then it is rated as high (H).(ii) If it supports two domains, then it is rated as medium (M).(iii) If it supports one domain, then it is called as low (L).(iv) If it supports no domain, then it is said as not applicable (N).
ese metrics are verified on different aspects such as service aspects, context aspects, and device aspects.

Service Aspects.
Service-oriented device middleware and service-oriented context middleware are two different middleware available to provide service aspects in the perspective of device aspects and context aspects, respectively.
is leads the researcher to move towards a middleware to manage both device and context functionalities in a particular middleware.Existing middleware are developed to provide service for the specific domain application, and there are no provisions to alter the available service to suit or accommodate to a different domain application.
Considering an example to illustrate how a same service works differently in diverse domain applications, consider the tracking as a service and it is applied in different domain applications, such as in the following: (i) In Scenario 1, an application of the transport system, location is tracked using GPS or some other techniques.Here, location is the context, and GPS is a device used to trace the location [33].(ii) In Scenario 2, an application of hospital, location of the doctor, patient, and object can be traced using RFID tag.Here, location, users, and objects are context, and RFID acts as a device [35].(iii) In Scenario 3, an application of home, location of the user/object can be identified using RFID.Mobile Information Systems Here, location, users, and objects are considered as context, and RFID belongs to device aspects [36].(iv) In Scenario 4, an application of supply chain management, transporting of goods from one location to another can be monitored using RFID and GPS techniques.Here, goods, location, and so on act as a context, and RFID and GPS act as some devices [37].(v) In Scenario 5, an application of the security system, movement of users and abnormal activities will be monitored using camera and identi cation sensors.
Here, movement of users and abnormal activities are considered as context, and camera and sensors act as devices [38].
From the above scenarios, it is evident that di erent tracking services are used for each pervasive application.Even though tracking is a common service, it varies only with the domain-speci c features, and also, there is no architecture or framework or middleware to make the service common to all domain applications as a generic service.
In service aspects, there exist ve middleware.ey are OASIS [39], SAMI [40], SOCAM [41], NEXUS [42], and KASOM [43] as shown in Table 2.Each middleware works  Mobile Information Systems for a specific application like sharing of information among devices in smart home, interface for network service in smart office, and so on.us, all the service aspects of context-aware middleware support for a specific domain application.

NEXUS.
It finds a new way to merge service-oriented architecture and pervasive computing to create a smart, healthy, and flexible network to access different active service resources [42].

SOCAM.
Pervasive application can be developed using service-oriented context-aware middleware platform.It mainly works with two layers.
e first layer is used to control and to manage context information, whereas the second layer helps to frame different policies to support various client applications [41].

OASIS.
It is object-centric service-oriented middleware to support different sensor devices, and it works according to the need of the service.It also supports automatic discovery of the service and the registry of service [39].

SAMI.
It is a self-adaptive information sharing middleware to locate different devices, and it is used for sharing the device information despite its dynamic location.Device information files are grouped as data [40].

KASOM.
e knowledge-aware and service-oriented middleware (KASOM) was illustrated as a new idea to combine pervasive computing and service-oriented cloud computing to create a novel era in the computing generation.
e major purpose of knowledge-aware and service-oriented middleware is to provide sophisticated and enhanced pervasive services to everyone connected to the smart environment.
From the above discussion, to support the diverse domain applications, the different domain services are to be maintained, and they should be provisioned to discover the suitable service and also to maintain the hierarchy of service [43].

Context Aspects.
Context-aware applications acclimatize according to the context of the user, context of the location, nearby people in the environment, objects, and the reachable devices in the environment, and changes take place to those objects over a period of time.A context-aware application with these potentials will analyze the environment and adapt according to the context of the environment [44].
Context of the application is monitored and processed with the support of smart devices, which acquires context information of the environment or its surrounding.In order to make context acquisition in reality is a demanding task.In order to achieve this, context aspect of the middleware is used, and it undergoes different challenges.ey are as follows [45]: (i) What are the different contexts that are available in general?
ere are different types of context available like the location, user, device, time, and activity [24].Context cannot be fixed in general, but it can be fixed for a specific application.
(ii) From where context is discovered?
Context information is taken from the available environment of the application like the location, user, and time and is monitored with the help of smart devices.
(iii) How to process the context information?
Context is processed generally in four different ways [7].ey are Same contexts will change from one domain application to another.For example, consider a location context which changes according to the domain.In smart hospital application, location contexts will be the doctor's room, ICU, general ward, and so on.In smart bank application, location contexts will refer to the manager's room, locker room, counter, and so on.In smart home application, location contexts will include bedroom, kitchen, hall, and so on.
In the context aspects, there exist twenty-four middleware, to mention few such as AURA [47], CARMEN [48], and CARISMA [49] as shown in Table 3.Here, different context-aware middleware work in different domains like smart home, smart hospital, and smart office.Each contextaware middleware works for a specific application.AURA is a context-aware middleware to design and develop a system to work as "personal information atmosphere" regardless of their location, and it acquires support of wearable computing, handheld device, sensor, desktop computers, users, and infrastructure computers [47].

GAIA.
GAIA is a distributed middleware system and acts as a metaoperating system to coordinate the different types of devices in the pervasive environment.As an operating system, it allocates resources, manages different file systems, and establishes communication with different resources [50].

MIPEG.
MIPEG is a middleware for pervasive grid applications, and it supports for integration of various mobile devices at anytime.It extracts the context information from the devices and executes the task according to the user task [55].

CAMPS.
CAMPS is an agent-based middleware to support context-aware service in the smart environment.Its primary objective is creating a standard to represent the context information and the method to process it [56].

5.2.16.
AMiCA.Ambient middleware for context awareness is used to continuously and implicitly adapt to the environment to meet evolving user expectations.Up-to-date valid context information is the key requirement for successful transparent interaction between the smart environment and the middleware [27].
5.2.17.IBICOOP.IBICOOP middleware is developed to support many challenges like abstraction of device configuration, connectivity of devices, and heterogeneity of devices [57].

CDTOM.
Context-driven task-oriented middleware focuses on the user's work or task in different situations, rather than focusing on the various devices and services in the environment.It also makes the system to work intelligently as per the user request [58].

UBIROAD.
UBIROAD is a middleware developed to handle the smart road environment.Here, context information is acquired from the vehicle and road in which the Mobile Information Systems vehicle is travelling (road condition, traffic, another vehicle detail, etc.) to create a smart environment for the user [21].
5.2.20.UMOVE.UMOVE middleware will consider every device as an entity, and also, it works based on the entitylayered architecture.It collects the information from the entity as entity data and processes the context of the entity data according to the application [59].

SMEPP.
SMEPP is a middleware to support security in the pervasive environment.It creates various security policies to access the context information which is acquired from the smart environment.According to the policy, middleware ensures the system is secured [21].

5.2.22.
Cameo.CAMEO middleware works for mobile devices and its context.It collects the context from the environment and analyze the context information.It provides a common programming interface to connect the smart environment.CAMEO middleware architecture also supports nonfunctional qualities like integrability, reusability, and ease of creation [60].

AWARE.
AWARE middleware is developed to collect the context from the environment.Accumulated context is managed and shared.It supports to present the context information to the stakeholders of the application.AWARE middleware architecture also supports nonfunctional qualities like scalability, integrability, reusability, and ease of creation.Most of the context-aware middleware work for the context representation such as the key-value model, logicbased model and ontology-based model, context acquisition, and context processing such as logical reasoning and probabilistic reasoning.ere is no middleware to support different domain applications in terms of all the aspects.e main objectives of the existing context-aware middleware are to investigate how contexts are acquired, processed, and modeled for a specific domain application.From the existing context aspect middleware, it is evident that there is no generic context-aware middleware to support context aspects.erefore, the need for the generic context-aware middleware is reported [61].

Device Aspects.
Currently, middleware is available to support specifically the device aspects of the domain which help in device management of a specific application, and in few middleware, generic device aspects are also supported.Hydra [62] and SenseWrap [63] are the middleware to support diverse domain applications, and both these middleware concentrate on (i) how to interface the heterogeneous physical device with the application software; (ii) how to handle the heterogeneous network protocol; (iii) device management.
It provides a device platform for context-aware middleware; despite this, it does not support context aspects.It uses virtual entity to map the physical devices.e virtual entity interfaces the physical devices to maintain the device properties.Any changes observed in the physical devices will be reflected in the virtual entity.ese mappings of the virtual entity help to achieve abstraction in the device management aspect.
Scope for interfacing the context aspect with the device aspect is shown in the study of Lowe.R et al. (2013).Both device aspects and context aspects combine together to provide a domain service to the user.
In device aspects, there exist twenty middleware such as HYDRA [21], MADAM [64], and PERLA [65], as shown in Table 4.Each middleware works for a specific application like interfacing the different devices in the specific environment of a domain.
Different middleware work in different domains like smart home, smart hospital, and smart office.Device aspect acts as a bridge between the physical device and context aspect of the system.It manages the devices in the smart environment to coordinate and act as a group to solve the different tasks of the application.

PICO.
PICO is developed to hide the complexity of integrating the heterogeneous devices of the smart environment.Its objective is to provide device-integrated service to any place at anytime.It is developed to support hospitalbased application [66].

HOMEROS.
HOMEROS middleware coordinates different devices with different properties in the pervasive smart environment.It provides the right service at right time to facilitate the context-aware application [67].

M-ECHO.
M-ECHO is used to support performance of the system even when available devices are connected through a weak network.It also adapts the morphing step to tune the peer-to-peer performance of the system and focuses on the data that are acquired [68].

MADAM.
MADAM mainly works to address the two issues such as to support mobile clients and heterogeneous devices.It also provides support to heterogeneity for both the client and the server.In the client, different context structures are framed to support different users, and in the server, specific services are created to support different requests [64].5.3.5.GAS-OS.GAS-OS is a middleware to separate the application logic from the complex pervasive environment by applying abstraction on the working of the heterogeneous device and environment [69].

GSN.
GSN is a middleware to provide virtual devices as a simple and high-level abstraction.It uses an XML representation to represent the various devices and is also used to connect the devices in the smart environment [70].

SATIN.
SATIN is a component-based middleware developed to support various mobile computing devices to handle the dynamic scenario in the smart environment [71].

MUNDOCORE.
MUNDOCORE is designed to support the communication between the networks and devices.It provides the platform-independent environment to support heterogeneous devices [72].
5.3.9.SAMI.SAMI is a self-adaptive information sharing middleware which works in a dynamic pervasive computing environment.It is used to share the information among the devices regardless of their network and environment [40].

ATLAS.
ATLAS is a sensor and actuator middleware platform to support integration of devices in the pervasive smart environment.Its main challenge is to handle heterogeneous devices and make them to work together to support different pervasive applications [73].
5.3.11.SANDMAN.SANDMAN middleware is developed to reduce the energy used in the smart environment by means of using the efficient protocol stack for communication.It also deactivates the unused devices to sleep mode which helps to reduce the power consumption [74].

PANOPLY.
e main objective of PANOPLY is to group the devices in the smart environment to solve the particular task according to the context of the pervasive smart environments [75].

MISSA.
e primary objective of MISSA is to separate the service logic from the data stream.Data stream is given more importance for the collection of context information from various devices and the proper standard for processing it [76].5.3.14.UBISOAP.UBISOAP middleware supports ubiquitous (anytime services) device services in the pervasive smart environment.It is achieved only by making the devices as a service provider and a service receiver [77].

ISMB.
ISMB is a middleware to support communication between devices and networks that can maintain one-to-one, one-to-many, and many-to-many relationships.It automatically discovers the device in the environment and also information maintenance of the mobile device [21].

HYDRA.
HYDRA middleware supports its application by allowing heterogeneous physical devices to flexibly adapt into their environments.It manages the devices in the environment and considers each device as a separate entity [21].5.3.17.SIRENA.SIRENA is mainly developed for the integration of smart devices in different domain applications.Devices are seamlessly connected in the smart environment.It is demonstrated with two domains, namely, industries and smart homes [21].5.3.18.ASPIRE.ASPIRE is a middleware to support tracking application along with many RFID devices and tools.It helps to track the number of goods in the environment [21].

PERLA.
PERLA is developed to provide support for managing and controlling the heterogeneous smart devices in the pervasive environment.It has its own language to describe the specification of the smart devices used in the application [65].

PECES. Pervasive Computing in Embedded Systems
(PECES) project is developed to create a framework to combine different devices of the smart environment to cooperate and work together to solve a particular task.It is demonstrated with the help of smart home application.
Most of the context-aware middleware work for the device management, and establishes communication between the various devices and middleware, and its focus is on device aspects.ere is no middleware to support different domain applications in terms of all the aspects [78].
Device aspects provide support to the physical layer and start with how to connect different devices without any problem and to maintain an active observer/listener to monitor the device and its changing information/data.Device data/information will be represented in a common way and stored in a common format for generic sharing of information.Devices are categorized as a group in order to interact with each group and solve a task effectively and smartly.
In order to support device aspects, a separate list of live devices is to be maintained, to retrieve and control the device information.e live list will give the exact number of devices working in the environment along with its properties.Different devices are grouped to achieve a common goal, and also, it should be ready to handle many requirements at the same time.

Existing Tool-Based Development Efforts
In order to test the working of the application model/ architecture, a tool-based method is needed for the developer.
e development tool acts as a test bed for most of the application developers.In recent times, pervasive application developers have started using various pervasive developer tools for their applications.e Context Toolkit [79] model is made up of different context widgets, and it is placed in a distributed environment.ese context widgets are used to access the context information and also hide the context-12 Mobile Information Systems sensing procedure.Context widgets are software components that are used for developing pervasive application.is model deals only with the context-sensing aspect.e tool Olympus [80] is to specify the devices, locations, and user in a high-level description form which in turn is converted into actual active space components that are used to build programs.Since this model specifies the context in a high-level description, it is difficult for the developer to understand.It is domain specific, so there are more difficulties in defining the specification of the application.
Archface is a tool [81] that acts as an interface between the architecture and methods and to implement the application.It has its own architecture description language to describe the specification of application, and also, it has programming level interface to convert the specifications into the implementation.e major drawback of the tool is that it deals only with architecture perspective and not with the various components that are used to develop the application.
PervMl [82] is a domain-specific language tool used to define the requirement specification in the metamodel design.is metamodel is in turn mapped to convert it into program that hides the high-level complexities involved in the development of the pervasive application.is model defines the specifications of the tool in a descriptive form, which makes the developer difficult for defining the specifications without predefined knowledge about the domain.
Diasuite [83] is a tool that allows the designer to define the taxonomy of specific application environment by using the existing application model.It has its own simulation tool to simulate the working of the application with the prewritten specification of the application.ese tools are taken as reference for the tool-based development.

Limitations of the Existing Work
From the observation of the literature survey, context-aware middleware is working on three different aspects as shown in Figure 5. e first aspect of the context-aware middleware concerns on domain on which it is operating, such as smart home, smart office, smart class, and smart hospital.e second aspect of the context-aware middleware focuses on the various devices operating on the pervasive smart environment and devices involved are dynamic or fixed to what extend is the great challenge.
e third aspect of the context-aware middleware deals with the context of the diverse domain environment.Context predicts the situation of the pervasive smart environment and adapts according to it.e different contexts are the location, time, device, operation, and so on, and the main objective is to assess how many contexts can the application support.
From the study, it may be observed that there are mainly three reasons to state middleware is specific to the application: (1) Context-aware middleware architecture is specific to the pervasive applications.ere is no generic context-aware middleware architecture to support diverse domain applications.
(2) Existing context-aware middleware implicitly states the limited nonfunctional qualities.From the literature survey, existing context-aware middleware architecture exhibits the limited number of nonfunctional qualities, and also, they are implicitly stated.(3) Context-aware middleware is biased to the specific aspects.Most of the existing context-aware middleware are biased of specific aspects like service aspect alone or only to context aspect or particularly to the device aspect.

Conclusions
e evolution of pervasive computing is presented.Various pervasive applications are obtained from the literature and classified according to their domain.
e different architectures for pervasive computing are analyzed, and it is found that context-aware middleware architecture provides a better solution to build pervasive application.In order to better understand the demands of stakeholders of contextaware middleware, a set of different aspects of context-aware middleware are derived from the literature and presented.
Context-aware middleware is classified based on three aspects such as context aspect, device aspect, and service aspect.Classification of the context-aware middleware has been given to obtain a clear understanding about the requirements of the different aspects.On analyzing the requirements of the context-aware middleware, the limitations are obtained, and the scope for the development of generic context-aware middleware is reported.

4. 2 .
Context Server Architecture.Context server architecture supports many client systems to access the remote data server concurrently.Data acquired are stored in the context server, and they are used according to the context of the application and contexts are monitored using di erent devices.Most of the devices used in context-aware systems are portable gadgets which are limited to power energy, computation power, memory space, and so on.ese are vital features for a centralized context server.It will consider di erent issues like fault tolerance, quality of services, and scalability of dynamic devices.

Table 2 :
List of context-aware middleware on service aspects.
(a) context data history, (b) context data aggregation through logical reasoning and probabilistic reasoning, (c) context data filtering using both time-based filtering and change-based filtering, (d) context data security.(iv) How context is represented?Context information is represented in any one of the five different models [7].ey are (a) key-value model, (b) markup schema model, (c) object-oriented model, (d) logic-based model, (e) ontology-based model.(v) How the conflict in the context is handled?Addressing two contexts of the application at the same time is a crucial decision to be made.It can be addressed in different ways [46]: (a) Conflict resolution Policy (b) QOC parameter (c) Priority to the context (vi) How same contexts differ from one domain to another domain?

Table 3 :
List of context-aware middleware on context aspects.
[48].CARISMA.CARISMA uses policies to work under dynamic context of the environment, and it adapts to the need of the application.Conflicts in the policies are solved during run time[49].Metadata help to find the dynamic context of the application and work according to it[48].
[52]5.COBRA.COBRA is an agent-based context-aware middleware to support context-aware application in the pervasive environment.It is the responsibility of the agent to collect the context information from the devices, environments, and other agents[52].givensystem,how to discover the resources available in an environment, and how to handle a resource failure and provide solution for it (Moushumi Sharmin et al., 2006).5.2.10.CARMEN.CARMEN is a middleware for handling context-aware resources in a wireless network, and it supports to create resource metadata.