Maruti Gupta

Energy impact of emerging mobile internet applications on LTE networks: issues and solutions

Mobile Internet applications run on devices such as smartphones and tablets, and have dramatically changed the landscape of applicationgenerated network traffic. The potent combination of millions of such applications and the instant accessibility of high-speed Internet on mobile devices through 3G and now LTE technology has also changed how users themselves interact with the Internet. Specifically, the radio states in LTE such as RRC_Connected and RRC_Idle were designed with more traditional applications such as web browsing and FTP in mind. These traditional applications typically generated traffic only during Active (Connected) state, and once the user session ended, usually the traffic ended too, thus allowing the radio to move to Inactive (Idle) state. However, newer applications such as Facebook and Twitter generate a constant stream of autonomous and/or user generated traffic at all times, thus erasing the previously clear demarcation between Active and Inactive states. This means a given mobile device (or user equipment, in LTE parlance) often ends up moving between Connected and Idle states frequently to send mostly short bursts of data, draining device battery and causing excessive signaling overhead in LTE networks. This problem has grown and attracted the research communitys attention to address the negative effects of frequent back and forth transitions between LTE radio states. In this article, we first explore the traffic characteristics of these emerging mobile Internet applications and how they differ from more traditional applications. We investigate their impact on LTE device power and air interface signaling. We then present a survey of state-of-the-art solutions proposed in literature to address the problems, and analyze their merits and demerits. Lastly, we discuss the solutions adopted by 3GPP including the latest developments in Release 11 to handle these issues, and present potential future research directions in this field.

Smartphone Background Activities in the Wild: Origin, Energy Drain, and Optimization

As new iterations of more powerful and better connected smartphones emerge, their limited battery life remains a leading factor adversely affecting the mobile experience of millions of smartphone users. While it is well-known that many apps can drain battery even while running in background, there has not been any study that quantifies the extent and severity of such background energy drain for users in the wild. To extend battery life, various new features are being incorporated within the phone, one of them being preventing applications from running in background, i.e., when the screen is off, but their impact is largely unknown. This paper makes several contributions. First, we present a large-scale measurement study that performs an in-depth analysis of the activities of various apps running in background on thousands of phones in the wild. Second, we quantify the amount of battery drain by all such background activities and possible energy saving. Third, we develop a metric to measure the usefulness of background activities that is personalized to each user. Finally, we present a system called HUSH (screen-off optimizer) that monitors the metric online and automatically identifies and suppresses background activities during screen-off periods that are not useful to the user experience. In doing so, our proposed HUSH saves screen-off energy of smartphones by 15.7% on average while incurring minimal impact on the user experience with the apps.

Power Saving mechanisms for M2M communication over LTE networks

Machine to Machine (M2M) communication is expected to be a major driver of growth in mobile communications for cellular networks such as LTE. Since LTE networks are primarily designed and optimized for human to human (H2H) communications, existing protocols and mechanisms are not very efficient for supporting M2M communication. Several modifications are needed to address the service requirements and traffic characteristics of M2M devices in these networks. Device power efficiency is one of the crucial requirements for M2M communication. We focus on solutions to decrease power consumption of M2M devices over 4G networks. M2M traffic characteristics are different from those of H2H traffic in terms of the size and the frequency of the generated data. One possible mechanism is to maximize the time that the M2M device is in low power state. In this paper, we evaluate the impact of extending the Paging Cycle and reducing the RRC Connected-to-Idle transition tail time on power savings. Our results show that for infrequent data transmission extending Paging Cycle reduces power consumption up to 79.3%. However, for frequent data transmission reducing Connected-to-Idle transition tail time is more effective and reduces power consumption by up to 76.2%.

Analyzing mobile applications and power consumption on smartphone over LTE network

The explosive number of applications now available on smart phone type of devices and the availability of high-speed mobile broadband connectivity through 4G networks such as LTE has only increased the challenges in improving battery life of these devices. In this paper, we illustrate the 4G power consumption issues through traffic traces and power measurements on LTE device over commercial LTE network. We analyzed the traces of various applications to understand the traffic characteristics. We also measured the power consumption of the various components of the device. This paper provides the cdfs of the packet size and packet inter-arrival times of background traffic of common applications (e.g., Skype, Google Talk, Twitter etc), and power consumption values and duration of low-power states which are crucial for the innovations and implementation of efficient power optimization mechanisms in LTE networks.

On the impact of small cell discovery mechanisms on device power consumption over LTE networks

Small Cells are under extensive investigation as a potential solution to meet the increasing capacity demand due to ever growing data traffic over cellular networks. The architecture of small cells is still under discussion, various architectures have been proposed based on potential use cases. One of the most important use cases is to deploy small cells on a different frequency than that of the macro cell to offload traffic. Small cell discovery in this type of inter-frequency scenario is an open research problem. Aggressive scanning mechanisms may result in faster discovery of small cells and higher user throughput as well as higher macro cell capacity due to potentially increased offloading opportunity. However, frequent scanning may also result in higher power consumption for the device. Therefore, a tradeoff between offloading opportunity and device power consumption seems inevitable in practice. In this paper, we analyze the trade-off between the offloading capacity and device power consumption due to inter-frequency scanning mechanism for small cell discovery. Our findings show that although increasing the frequency of inter-frequency scanning does result in increased power consumption due to scanning, it actually lowers the total power consumed by the device since it leads to offloading to small cells, where the device consumes less overall power due to better coverage.

A Fine-grained Event-based Modem Power Model for Enabling In-depth Modem Energy Drain Analysis

Cellular modems enable ubiquitous Internet connectivities to modern smartphones, but in doing so they become a major contributor to the smartphone energy drain. Understanding modem energy drain requires a detailed power model. The prior art, an RRC-state based power model, was developed primarily to model the modem energy drain of application data transfer. As such, it serves well its original purpose, but is insufficient to study detailed modem behavior, eg. activities in the control plane. In [2], we propose a new methodology of modeling modem power draw behavior at the event-granularity, and develop to our knowledge the first fine-grained modem power model that captures the power draw of all LTE modem radio-on events in different RRC modes. Second, we quantitatively demonstrate the advantages of the new model over the state-based power model under a wide variety of context via controlled experiments. Finally, using our fine-grained modem power model, we perform the first detailed modem energy drain in-the-wild study involving 12 Nexus 6 phones under normal usage by 12 volunteers spanning a total of 348 days. Our study provides the first quantitative analysis of energy drain due to modem control activities in the wild and exposes their correlation with context such as location and user mobility. In this abstracts, we introduce the essence of the methodology and the highlighted results from the in-the-wild study.