Tendencia a la armonización internacional: hacia un Derecho mercantil

The impact of mobile traffic offloading can be evaluated from two aspects: the operator perspective and the user perspective. The ultimate goal is to benefit both network operators and users. However in reality, the main concern of network operators is about how to offload as much traffic as possible to alleviate the pressure on their network infrastructure, whereas a key factor from the user perspective is the battery life. Although much of the prior work in traffic offloading focused on the operator perspective, recent studies have looked into the user perspective and shown that by improving energy awareness in traffic offloading we can extend the battery life for mobile users and still provide a desirable outcome from the operator perspective [15, 16].

To get a better understanding of energy cost in mobile traffic offloading, we analyze the offloading procedure and further suggest viable approaches as highlighted in Table 2.1 to enhance energy awareness.

• In the initiation phase, if offloading is triggered by the network side, signaling consumes energy. If signaling messages are delivered too frequently with a large volume of data, such unintentionally recurrent interactions may promote the cellular interface state, thereby leading to excessive draining of the battery [131]. This problem is comparable to the case of background traffic. On the other hand, the user-driven initiation can not benefit from the proactive guidance but it does not consume extra energy in this phase.

Based on the observation, it is crucial to avoid frequent network sig- naling with large payload, which can lead to extra energy consump- tion due to the change of cellular radio state. In order to benefit from the network support and to support dynamic traffic management, an adaptive scheme that combines the user-driven and network-driven initiation is recommended. Such a scheme helps strike a balance be- tween the efficiency and energy consumption [16, 92].

• The context collection phase has to be done carefully. As revealed in a previous study, frequent scanning of available WiFi networks consumes a significant amount of energy [92]. If GPS is used, a cold- start can take around 20 seconds and consume ∼6.3 Joules [16]. If the context information is delivered to a remote server, it also costs energy.

In the context collection phase, constant scanning must be avoided if possible for its high energy consumption on mobile devices. Due

to the relatively high energy consumption and the long latency to obtain a fix from the cold-start, the GPS usage in traffic offloading needs to be assisted by energy efficient design, such as using tech- niques proposed in [210, 211]. To support energy efficient position- ing, WiFi-based positioning can also be used as an alternative. For context processing and management, it is recommended to share the load between local devices and remote servers, which can fully utilize the knowledge of access network and infrastructure, as well as the computing resources of both sides.

• In the decision making phase, if the computation is done locally,

Table 2.1: Approaches to enable energy-aware traffic offloading.

Phase Recommendations

Offloading 1) Avoid frequent signaling from the network side that Initiation triggers cellular state change.

2) Combine the user-driven and network-driven in an adaptive manner to improve efficiency.

Context 1) Avoid unnecessary scanning and GPS operation. Collection 2) Utilize an energy efficient positioning mechanism such

as WiFi fingerprinting.

3) Adapt the context management for both local and remote processing to strike a balance.

Offloading 1) Utilize energy-aware algorithms to guide the decision. Decision 2) Adopt dynamic mechanisms to update the logic

according to the network condition.

3) Utilize the cloud support to offload the energy cost from intensive computation.

Network 1) Avoid time-consuming association operation or protocol. Association 2) Utilize guidance from the network side if possible to

assist authentication and access connectivity.

Data 1) Adopt optimization schemes for different types of traffic. Transmission 2) Avoid transmission over unstable or low throughput

wireless links by predicting the user mobility and network condition.

Offloading 1) Utilize hints if possible from either network side or local Termination controller for switching between connections.

2) Avoid frequent terminations that can shorten the data transmission time.

the computational complexity of the algorithm determines the en- ergy consumption. If the computation is offloaded to remote servers, the data communication due to the exchange of context and decision messages will consume energy.

To enable energy awareness, energy consumption must be a key fac- tor in making traffic offloading decisions. In this phase, a dedicated energy-aware algorithm can be used together with other factors such as network performance and offloading capacity. Due to the fast change of network conditions, the energy-aware offloading decision needs to be adaptive to adjust its operation logic. To save energy consumed by computation, it is recommended to offload intensive computational loads such as mobility prediction to the network side. • In the association phase, the connection establishment that involves various hardware and protocol operations can consume a non-negligible amount of energy. For instance, DHCP alone can consume 4.8 Joules [16]. Therefore, when the user is moving at a high speed, the frequent associations accelerate the battery drain.

In order to minimize the time spent in the association phase, it is rec- ommended to use light-weight configuration protocols and to avoid the ones requiring complex message exchange. To assist authentica- tion, network support can be utilized to deliver configuration infor- mation.

• In the data transmission phase, the amount of available bandwidth and wireless condition greatly affect the energy consumption. By taking WiFi offloading as an example, when WiFi signal strength is poor, offloading cellular traffic to WiFi can cost more energy [16]. For data transmission, the first recommendation is to adopt opti- mization techniques tailored for characteristics of different types of traffic and thus maximizing the transmission throughput during the offloading period. For instance, data batching and energy efficient scheduling of delay tolerant traffic can effectively amortize the penalty caused by the tail energy [130]. Several optimization techniques related to this have been proposed in the literature [5, 190, 191, 192]. The second recommendation is to avoid using unstable or low throughput wireless connections by estimating the network condition and user mobility [85, 86, 87, 124], [212]–[216].

• The termination phase should be handled in a timely manner. Oth- erwise more energy can be consumed during the periods of poor con-

nectivity and low bandwidth.

Since the offloading termination affects the data transmission and ser- vice quality, we can improve energy saving by utilizing proactive ap- proaches to plan the termination and apply efficient handover schemes to ensure the service continuity. There has been plenty of work on this [55], [63]–[68], [81]–[84]. To avoid frequent unnecessary terminations that degrade the transmission performance, it is recommended to utilize the knowledge from the network side (e.g., network setup and position of APs) combined with the mobility prediction techniques. Mobile traffic offloading provides a promising way to alleviate the pres- sure on existing cellular networks overloaded by data traffic and it is gaining support in the future due to the rapid growth of mobile data traffic. By making the offloading process energy aware, it is possible to improve user experience by extending the battery life on mobile systems.