Throx

Author: T. Y. Fam
Research Assignment

Energy-Efficient mobile computing approaches

1. Introduction

1.1 General Information
Mobile or wireless devices have become important tools for the modern world. For example GSM mobile, ad hoc networks and sensor networks are good examples of modern wireless networks. However the current battery technology couldn’t catch up with the implementation of mobile wireless technology. Mobile devices are always stuck with the limitation of battery life. Therefore a good protocol has to be implemented to overcome this matter. In mobile devices, the main consumption of battery power is used by the circuit board of the device. Then the second consumption is used for transmitting and receiving the total traffic. The third consumption will be the output of the devices [1].

Capacity of battery is usually based on the size of the battery itself. Obviously, the bigger the battery, the longer the battery will last. Big-sized battery is not the solution for mobile devices, as mobility required small, light and portable batteries. Devices that operated on AA size batteries should focus on the battery issues. The battery life should last at least 2 to 3 hours before the next recharge cycle. Battery experts make a prediction that battery technology will only improve 10 to 30% in the next 5 to 10 years [3].

Currently new mobiles devices are all designed in a downsizing manner that can fit in our palm such as Pocket PC and PDA Palm. In the future, if battery power problem can be solved, they all can establish a reliable distributed network that can serve people well. The demand of portable file system keeps increasing as more people wants to access their file on the move. To achieve that, battery power problem has to be solved to provide a reliable and constant connection to the networks. Unreliable battery power will cause disconnection and poor network latency performance such as limited bandwidth, because the device’s wireless interface may need to be turned off to recover or conserve its battery energy [3].

To reduce energy consumption in mobile networks is a crucial and an important task. Hardware has been improving in recent years for the aim of achieving low energy consumption standard. However due to the slow improvement of battery technology, others area of possibility of energy consumption method has to be made in order to cooperate in conserving energy. Due to the limitation of hardware design, energy consumption has to be done in protocol design. A protocol that uses schedule mechanism will allow the wireless interface enough time to turn off, thus allowing the battery cells to regenerate its energy.

2. Infrastructure and Ad-hoc networks

2.1 Infrastructure Network
There are two normal wireless architectures being used in our current wireless network system. They are infrastructure and ad-hoc networks. Infrastructure networks are often connected to a base station such as desktop or server that is installed with different wired adapters. They work as a router to connect the device to other devices wirelessly through different channels [2]. Those channels can be FDMA (Frequency Division Multiple Access), TDMA (Time Division Multiple Access) or CDMA (Code Division Multiple Access) which are also widely used in phone networks. All the server or host know how to communicate with each other on different channels, thus they will act as the last route for mobile devices. An example of infrastructure network can be seen in Figure 1.

2.2 Ad-hoc Network
Ad-hoc networks are a multi-connection network of mobile devices itself without the need of a server to route the communication channel. This type of network will be completely devoid of wires, all connections are made with wireless. An idea of Ad-hoc networks is shown in figure 2. All wireless devices communicate with each other rather than connecting to a server first. This method is a good temporary network environment especially for military needs. All the devices must be able to exchange topology information in order to route the networks successfully [2].

3. Power Consumptions

3.1 General Information
Software or applications in mobile devices do not interact with the wireless network directly, in fact they communicate with the protocol software and then the protocol software interact with the hardware to communicate with the networks. Efficient energy research has been done around the physical layer because people believe that the power consumption of a device is mainly by the system hardware itself [2]. There are many researches that tend to increase the battery life by increasing the battery size but bear with the limitation that we cannot have a bulky battery in our mobile devices such as cell phone or PDA. Battery technology has not been improving much for over 30 years, unlike other part of computer system such as CPU, the technology keep changing every year. The usual way to reduce battery consumption is to reduce the processing speed and hard disk spinning speed [2]. All these ways are able to reduce the battery consumption significantly but more ways should be considered to improve the battery life. One of the ways would be the refinement of the current protocol layers.

A basic wireless communication devices required receiving, transmitting and standby mode. Transmitting required the most power consumption and the least for standby mode [2]. For example, power consumption for Proxim RangeLAN2 2.4GHz PCMCIA card needs 1.5Wolt for transmitting, 0.75Wolt when receiving and 0.01Wolt while idle [2]. Therefore, the way to reduce energy consumption would be the implementation of new energy savings protocol to reduce the transmitting power.

4. Energy-efficient protocol

4.1 General Information
Before we can start discussing the way of implementing an energy-efficient protocol, we need to understand what the characteristics of current wireless protocol are. For IEEE 802.11 wireless mechanism, the receiver is turned on at all times to receive every packet and check for the required packets. This mechanism will waste lots of battery energy as it keeps on monitoring all the packets. Solution can be made if they schedule the transmission time; receiver will switch off until the transmission time started. The other way is the receiver will assume that it does no need to receive any more packets and it will turn off the receiver [2]. To established a complete connection, device have to switch from transmit to receive mode and vice-versa, that will cost a lot of time and wastage of energy. This can be solved by allocating contiguous transmitting slots so that it can transmit or receiving in a longer time before each mode switching [2].

4.2 IEEE 802.11 Protocol
Three widely used protocol named IEEE 802.11, EC-MAC and PAMAS protocol are all designed with the idea of energy savings but not efficient enough for today’s mobile technology. IEEE 802.11 protocol basically was designed based on CSMA/CA (Collision Sense Multiple Access / Collision Avoidance). IEEE 802.11 uses a transmission channel to transmit packet. Mobile will capture the channel and transmit all data if the channel is not busy [2]. If the channel is busy, it will enter into the “off” state again. This protocol conserves power by switching the mobile to sleep mode and informing the station of this action. When there is data that needs to be sent to the sleeping mobile, the station will send a small buffered packet on air and wait for the mobile to wake up and receive it [2]. This method can conserve some energy but may result in high latency for the wireless network. Even though the IEEE 802.11 protocol can conserve energy, but energy efficient was not the main issue when designing this protocol compared to EC-MAC protocol [2].

4.3 EC-MAC Protocol
EC-MAC (Energy Conserving – Medium Access Control) protocol was designed based on energy-efficient as a main goal [2]. The EC-MAC protocol usage in infrastructure network is as mentioned before; where a base station routes the entire mobile client. Basically it uses reservation and scheduling methods to conserve energy [2]. EC-MAC protocol uses a frame with multiple data slots to transmit its data.

Every time EC-MAC transfers the frame, it will include a Frame Synchronization Message (FSM) at the beginning followed by request or update, new user, scheduled message and data phase [2]. In request or update phase, mobile will send a request of establishment of queues which has been set in the FSM. Collision is avoided by letting the base station control the order of reservation transmission within this phase. The new user phase will allow new user to register to the network. In schedules message phase, it will broadcast the slot permission for the transmitting data. Downlink and uplink are both used to store the data that needs to be transmitted [2].

EC-MAC protocol can reduce energy consumption because its uses a centralized scheduler as mentioned before [2]. This can avoid lots of collisions in a wireless environment. With the scheduler system, client mobile will not be required to keep scanning the whole channel for packets. Centralized scheduler will also provide contiguous slot allocations for each mobile device, this reduce the transmission mode changes that increases energy consumption. It can also assign appropriate priority to each client mobile based on their battery power level [2].

4.4 PAMAS Protocol
PAMAS protocol is similar with EC-MAC protocol that they both are designed as a main idea to conserve energy. PAMAS is designed mainly for ad-hoc network with the modification of MACA protocol. In PAMAS protocol, if a mobile wants to send a packet to the receiver mobile, it first send a RTS (request to send) signal to the receiver and wait for CTS (clear to send) signal [2]. If the receiver rejects the request, the sender will enter into “off” mode. The mobile that is receiving packets will send a “busy” signal to the channel to block further incoming request. Power conservation will be achieved by turning off the wireless receiver when they are not sending or receiving packets. With the busy signal, other will not be able to request RTS signal and thus conserve energy [2]. The main idea of using RTS and CTS is to determine when to turn off the wireless device to save energy. Simulation of analysis shows that around 10 to 70% of energy can be conserved with this protocol in the topology [2].

5. Battery Technology

5.1 General Information
It is important to understand how a battery works before we can discuss what the ways to utilize battery efficiency are. A battery consists of many cells that arranged in an array. Each cell contain anode and cathode which separated by electrolyte. Electric current is generated by the chemical reaction from the anode/cathode – electrolyte interface. As the use of current is increased, the cell voltage will be decreased and this is called “rate capacity effect” [1]. When the materials such as electrodes and electrolytes used to generate power is decreased, the cells voltage will significant drop down to an unusable level. Therefore the amount of electric that can be generated by a battery is based on the amount of electrodes and electrolytes the battery have [1].

Normal way of battery energy drain is by using a constant energy drain method. Using constant discharge method will drain battery life much faster because materials used in battery cells will not gain enough time to get compensation for the depletion. Even though there improvements have been made in battery technology, constant discharge will still only receive 10-30% power from a full charge battery [1]. The only way to solve this problem is using discharge pulsed. The idea is to allow the battery cells enough time to rest and regenerate for enough active materials to produce energy. By using discharge pulsed method, the energy delivered by a battery will be greater [1]. Therefore by using a battery discharge shaping method, it is obvious that it can increase the energy of battery efficiency.

5.2 Stochastic Emulation
A battery recovery model has been introduced in [1]. This model designed so that it uses a stochastic process to emulate the discharge of a battery. The model assumed that time scale is divided into time slots and one discharge pulsed represent by one time slot. This method was tested by the function of number of discharge per time slot that can be emulated by changing the value of number of discharge pulsed [1]. This model proves that energy can be increased if the number of discharge pulses is decreased per time slot, because the time of recovery of each slot is increased. It’s also shows that the more discharge pulses it have, the weaker the battery life would be.

5.3 Scheduling Scheme
Another approach was introduced in [1] is that the used of smart scheduling algorithm with discharge shaping. Battery will be divided into parallel of cells that can be selected specifically by the algorithm. Scheduling scheme will detect and choose the fully charged cell to drain energy. When the cell energy drop down to a specific threshold, it then switch to the next cell and continue the discharge process. This method will allow all the cells enough time to recover their battery life. The scheduler will always choose the battery cell that has the best state of charge [1]. When the energy of each cell drops down to a threshold, it will be removed from the active cell and allowed time for recovery. If none of the cells are available when the discharge arrives, the discharge process will be buffered and waits for the next available cell. By using this scheduler scheme, battery will be able to produce its maximum energy even with a continuous discharge requests.

6. Energy-efficient Protocols Design (Network Layer)

6.1 General Information
Network layer is used for routing packets in both infrastructure and ad-hoc network. However, the function of energy efficiency will not apply in infrastructure network because base station will do the routing job for all the client mobiles. But energy consumption plays an important role in ad-hoc network. As mentioned before, ad-hoc networks are a multi connection of mobile devices itself without the need of a server to route the communication channel. Ad-hoc networks are complicated because it required updating the network topology details every time [1]. Figure 4 represent an example of a small ad-hoc network, if A wants to communicate with C, the network should choose the shortest route which is through A, E and destination D. But it will continuous drain the mobile E’s battery life and will be forced to turn off. In the end it will cause F could not connect to the entire networks. Therefore, a good routing algorithm should be designed so that the route of A to D should not always use route E but using the route of B and C [1].

6.2 Unicast Traffic
There are two types of methods to achieve the above approach which are Unicast traffic and Broadcast traffic. Unicast traffic tends to conserve energy by choosing the cost effective routes rather than the shortest path [1]. This is to ensure that all the mobiles can stay alive as long as possible. Although packets that are transmitting through a longer path might logically drain more battery life, but it can actually conserve energy in some way. For example E may be a congested node, going through B and C will reduce the frequent changes of mode in E [1]. People argued that by choosing a longer path may result in high latency networks. But this is not often true because shortest path usually will be the highly congested path; therefore by using the longer path can actually have the same or better performance of short path routing [1].

6.3 Broadcast Traffic
Broadcast traffic simply broadcast packet to all the nodes, each mobile will receive the packet once and then turn off if they know theirs neighbors has already received the packet. Only the intermediate nodes will be required to retransmit the packet to next destination. The traditional broadcast way of broadcasting is simply flooding algorithm because it keeps sending the packet to all mobiles until it reaches the destination [1]. Therefore it is not suitable for wireless network as mobile devices will consume lots of energy if they keep on broadcasting the same packet.

Networks can be divided into different sets of trees. When a mobile want to broadcast a packet, it can choose the tree that have most power energy available. One of the way is to ensure that the packet arrive at as many new nodes as possible in order to expand the connection area. A packet that in a specific tree can travel to a different tree if the energy allows, priority will be given to a tree that generally have more energy power and more nodes available [1].

7. Application Layer

7.1 Video encoding and Proxies
Apart from the refinement of protocol layers, energy consumption reduction can be done in application layer as well. Advanced tools such as Power Monitor and PowerScope can help software developer to develop energy conserving applications. A report shows that 46% of energy reduction by a video player application by using the information and advantages provided by PowerScope [1]. Another way of conserving energy and bandwidth is to use proxies. It helps reduce the processing power of application such as browser and streaming software. Proxies can control the streaming content such as blocking the video and only allows the sound to be streamed to the mobile devices [1]. Proxies are usually installed at the base station that is located closest to the mobiles.

7.2 Multimedia Processing
Multimedia processing is deemed to be the highest energy-consuming task in mobile networks. Video processing and transmission often takes up a lot of energy and bandwidth due to the high data packet of video format. Therefore video encoding and decoding plays a main role in wireless mobile networks. There are two main methods to reduce the bits in video streaming. First, video can be encoded to a lower bit by using encoder such as DIVX then send to the receiver, the challenge is that the reduction of quality should be kept minimal. Receiver should also make sure that they have the appropriate decoder to play the video correctly. By reducing the video file size, transmission of the transmitter will be reduced thus reduces the energy consumption by the wireless interface. Secondly, remove selected bit from the network interface card. This can be achieved if the network card is sensitive to battery energy power status, it can then remove unwanted or unimportant packet during the video streaming transmission.

8. Conclusion
As conclusion, mobile wireless technology is changing and improving everyday. The main issue will be the battery technology that hasn’t been improved very much for the last 5 to 10 years. It can be deduced from the researches above, the longer the idle time of the transmitter, the longer the battery life that can be saved due to the nature of battery structure. The design of good energy-efficient mechanism is not only limited to protocol design, many other areas can be improved at different layers. Low power consumption protocols remain one of the main important tasks for the wireless technology’s researchers and communities. Designers must focus on designing energy-efficient protocol within the physical layer and application layer. A balanced must be achieved between delay and battery life, latency should be kept as low as possible while keeping a good battery life.

List of References

[1] Carla, F. C., Pavan, N., Vikram, S. (2002), “Energy-Efficient Communication Protocols”, ACM, New Orleans, Louisiana, USA, pp. 824-829

[2] Christine, E. J., Krishna, M. S., Prathima, A., Jyh, C. C. (2001), “A Survey of Energy Efficient Network Protocols for Wireless Networks”, Wireless Networks, Kluwer Academic Publishers, Netherlands, Vol. 7, pp.343-358

[3] Jun, C., Kian, L. T. (1999), “Energy-efficient selective cache invalidation”, Wireles Networks, J.C. Baltzer AG, Science Publishers, Vol. 5, pp.489-502

[4] Kian, L. T. (2001), “Organization of Invalidation Reports for Energy-Efficient Cache Invalidation in Mobile Environments”, Mobile Networks and Applications, Kluwer Academic Publishers, Netherlands, Vol. 6, pp.270-290