LINEA ESTRATÉGICA: MODELO DE ORGANIZACIÓN INSTITUCIONAL
MODELO DE ORGANIZACIÓN
6. CONCLUSIONES Y RECOMENDACIONES
Advancements in mobile device technologies have led to the production of more complicated and interactive applications such as face recognition, real-time video, healthcare monitoring, Augmented Reality, and Virtual Reality mobile applica- tions. The common feature between these applications is that they all need high processing power and fast and near real-time response time. Due to these require- ments, the concept of computation offloading from mobile devices to resourceful cloud servers has been introduced as MCC. The complicated computation-intensive modules of the mobile applications can be easily executed with a fast response time on the powerful remote surrogates [77]. However, the performance of offloading is highly affected by the available bandwidth and latency [129]. Mobile computation offloading will be explained in detail in Section 2.2.
MCC is an emerging paradigm that encompasses mobile computing, cloud computing, networking and virtualization. It is well studied in the literature. There are many extensive MCC surveys in the literature [44] [128] [38] [76]. MCC can support mobile devices through either computation augmentation which is about leveraging powerful external resources to execute the heavy parts of the application or storage augmentation for using the excessive data storage capabilities of the cloud [163].
Because of high latency and low bandwidth issues of remote cloud servers in traditional MCC studies, researchers have suggested using one-hop away powerful surrogates to the mobile devices in the form of Cloudlets which is discussed in Section 2.1.1.1. As smartphones and tablets gain more CPU power and longer battery life, the meaning of MCC gradually changes. Instead of being fully hooked to Remote Cloud, a number of Nearby Mobile Devices can be used to coordinate and distribute content in a decentralized manner discussed in Section 2.1.1.2.
2.1 Mobile Cloud Architectures 15 2.1.1.1 Cloudlet
Cloudlet is viewed as a small-scaled cloud having a cluster of computers that is well-connected to the Internet serving mobile devices in close proximity. In this way, we manage to overcome the high latency problem faced in the traditional MCC approaches, which depend on Remote Cloud servers only. The idea of using cloudlets (surrogates) as near fixed devices was introduced in [120] in which the mobile device offloads its workload to a local cloudlet comprised of several multi- core computers with connectivity to the remote cloud servers. Satyanarayanan [119] uses four futuristic scenarios as case studies for the facts on the architectural evolution of mobile cloud computing from a 2-tier (mobile device - cloud) to a 3-tier (mobile device - cloudlet - cloud). Several approaches to this design are mentioned in this paper, such as VM synthesis and code offloading. Alongside solving latency issues, he sheds light on other value propositions of cloudlets such as the bandwidth, crowd-sourcing, privacy and security, and availability.
From an energy point of view, offloading to cloudlet saves energy because of access to the short-range wireless connection [48]. Sending and receiving data not only claims large shares of wireless bandwidth but also drains the battery of mobile devices. It is shown that despite several new power-saving mechanisms of 4G/LTE cellular data; it requires 23 times more power than WiFi and it is less power efficient than 3G [61]. In some works, the resource-intensive application tasks are first offloaded to the cloudlet, and then later to nearby devices in the LAN or remote cloud resources if needed [147]. In case the connection to cloudlet service is interrupted, the offloaded tasks can be retrieved back and send them to be executed on alternative avenues.
The main goal of using cloudlets is to be able to reduce the response time in order to meet the needs of some latency-sensitive applications [49]. The advantage of being close to the user allows us to achieve this goal but does not provide us with the same computation and storage capabilities as those of the remote cloud. Thus, its computing capacities are limited to certain services. In addition to its proximity to the users, Cloudlet also has the advantage of being exploited by mobile users who do not even have an Internet connection [120].
2.1.1.2 Mobile Ad-hoc Clouds
The concept of Ad-hoc Cloud Computing has been discussed in [91]. According to [44], apart from offloading to central clouds, there are two other definitions of MCC. The first is where some mobile devices act as cloud resource providers forming a Peer-to-Peer (P2P) network. In this model, the mobile devices in the
local vicinity and other stationary devices (if available) would create an ad-hoc network which can be accessed by other mobile devices in order to run their applications. Theoretically, this model allows offloading the tasks to the mobile devices that form the virtual resource cloud. Besides, latency is also reduced, since the mobile users only have to access the virtual cloud resource instead of traversing lots of hops to get to the Remote Cloud.
MACs are about leveraging the computational capabilities of the surrounding mobile devices by having them as resource nodes. It is also known as Mobile Device Clouds (MDC) [95], Mobile Edge Clouds (MEC) [40] [14] [45], and Virtual Computing Provider for Mobile Devices [62]. The most well-known works at accomplishing resource sharing and data distribution among mobile devices in the vicinity include Hyrax [90], Scavenger [75], Serendipity [124], and Cirrus [122].