Energy-efficient wireless networking for multimedia
`applications
`(Wireless Communications and Mobile Computing, Wiley, 2001: 1:165-184)
`
`Paul J.M. Havinga, Gerard J.M. Smit
`University of Twente, department of Computer Science, Enschede, the Netherlands
` {havinga, smit}@cs.utwente nl
`
`
`
`
`Abstract – In this paper we identify the most prominent problems of wireless multimedia
`networking and present several state-of-the-art solutions with a focus on energy efficiency.
`Three key problems in networked wireless multimedia systems are 1) the need to maintain a
`minimum quality of service over time-varying channels, 2) to operate with limited energy
`resources, and 3) to operate in a heterogeneous environment. We identify two main
`principles to solve these problems. The first principle is that energy efficiency should
`involve all layers of the system. Second, Quality of Service is an essential mechanism for
`mobile multimedia systems not only to give users an adequate level of service, but also as a
`tool to achieve an energy-efficient system. Due to the dynamic wireless environment,
`adaptability of the system will be a key issue in achieving this.
`Keywords – energy efficiency, wireless networking, mobile computing, quality of service.
`
`1
`
`Introduction
`
`Advances in technology enable portable computers to be equipped with wireless interfaces, allowing
`networked communication even while on the move. Whereas today’s notebook computers and personal
`digital assistants (PDAs) are self contained (introvert) tomorrow’s networked mobile computers are part
`of a greater computing infrastructure (extrovert). Key problems are that these portable wireless network
`devices need to handle multimedia traffic in a dynamic and heterogeneous wireless environment, and the
`need to operate with limited energy resources.
`Wireless communication is much more difficult to achieve than wired communication because the
`surrounding environment interacts with the signal, blocking signal paths and introducing noise and
`echoes. As a result wireless connections have a lower quality than wired connections: lower bandwidth,
`less connection stability, higher error rates, and, moreover, a highly varying quality. They need to be able
`to operate in environments that may change drastically – in short term as well as in long term – in
`available resources and available services. These factors can in turn increase communication latency due
`to retransmissions, can give largely varying throughput, and incur a high energy consumption.
`Wireless networking is a broad area, and has many applications ranging from voice communication
`(cellular phones) to high performance multimedia networking. In this paper we are somewhat biased
`towards multimedia traffic, as it is expected that the new generation of wireless networks will carry
`diverse types of multimedia traffic.
`
`Key problems of wireless multimedia networking
`1.1
`Three key problems in wireless multimedia networking are 1) to operate with limited energy resources, 2)
`the need to maintain quality of service (throughput, delay, bit error rate, etc) over time-varying channels,
`and 3) to operate in a heterogeneous environment.
`Energy-efficiency – Portable wireless devices have severe constraints on the size, the energy
`consumption, and the communication bandwidth. Moreover, it is expected that these devices will be
`multimedia oriented, and need to handle many different classes of data traffic over a limited
`bandwidth wireless connection, including delay sensitive, real-time traffic such as speech and video.
`More extensive and continuous use of network services will only aggravate this problem since
`
`Smart Mobile Technologies LLC, Exhibit 2010
`Page 1 of 26
`
`

`

`
`
`communication consumes relatively much energy. Unfortunately, the rate at which battery
`performance improves (in terms of available energy per unit size or weight) is fairly slow, despite the
`great interest generated by the booming wireless business. Aside from major breakthroughs it is
`doubtful that significant reduction of battery size and weight can be expected in the near future. The
`energy consumption these devices need for communication and computation will limit the
`functionality of the mobiles.
`The way out is energy efficiency: doing more work with the same amount of energy. The art of low-
`power design used to be a narrow speciality in analog circuit design. Nowadays, it is appearing in
`many layers of a system.
`Quality of Service – A Quality of Service model provides the basis for modern high-bandwidth and real-
`time multimedia applications like teleteaching and video conferencing. The notion of QoS service
`originally stems from communication, but because of its potential in the allocation of all scarce
`resources, it has found its way into other domains, e.g. operating systems [25].
`Heterogeneity – In contrast to most stationary computers, mobile computers encounter heterogeneous
`network connections. As they leave the range of one network transceiver they switch to another. In
`different places they may experience different network qualities. There may be places where they can
`access multiple transceivers, or even may concurrently use wired access. The interface may also need
`to change access protocols for different networks, for example when switching from wireless LAN
`coverage in an office to cellular coverage in a city. This heterogeneity makes mobile computing more
`complex than traditional networking.
`These three problem-areas that are characteristic for future mobile networking are strongly correlated.
`[23]. Wireless network protocols typically address network performance metrics such as throughput,
`efficiency, fairness and packet delay. In this paper we address the additional issue of energy efficiency of
`the wireless network protocols 1. Considerations of energy efficiency are fundamentally influenced by the
`trade-off between energy consumption and achievable Quality of Service. The dynamic communication
`and application environment is an extra challenge, which might be solved using QoS principles. The aim
`in mobile multimedia networking is to meet the required QoS, while minimising the required amount of
`energy. To deal with the dynamic variations in networking and computing resources gracefully, both the
`mobile computing environment and the applications that operate in such an environment need to adapt
`their behaviour depending on the available resources including the batteries. Current research on several
`aspects of wireless networks (like error control, frame-length, access scheduling) indicate that continually
`adapting to the current condition of the wireless link have a substantial impact on the energy-efficiency of
`the system [8][10][33].
`
`Principles of energy-efficient wireless networking
`1.2
`In this paper we will discuss a variety of energy reduction approaches that can be used for building an
`energy-efficient mobile system, and show the relationship with multimedia and the dynamic environment.
`In this paper the following main principles are identified:
`1.
`Involve all layers. Energy efficiency is an issue involving all layers of the system, its physical layer,
`its communication protocol stack, its system architecture (Section 2.1), its operating system, and the
`entire network (Section 4).
`2. Quality of Service is an essential mechanism for mobile multimedia systems not only to give users an
`adequate level of service, but also as a tool to achieve an energy efficient system.
`QoS support in wireless networks involves several considerations beyond those addressed in earlier work
`on conventional wireline networks. In traditional networks, based on fixed terminals and high-
`quality/high-capacity links it is feasible to provide 'hard' QoS guarantees to users. However, in a mobile
`environment, mobility and the need for efficient resource utilisation require the use of a 'soft' QoS model
`[44]. The minimum QoS requirements for multimedia applications has a wide dynamic range depending
`on the user's quality expectations, application usage models, and application' tolerance to degradation.
`Users and applications require a certain QoS level. The system then operates in such a way that it will try
`
`
`1 In general, saving energy for the base station is not really an issue, as it is part of the fixed infrastructure and
`typically obtains energy from a mains outlet. However, since the current trend is to have ever smaller cell sizes, and
`the complexity of the base station is increasing, this issue might become more important in the future mainly because
`of economical and thermal reasons.
`
`
`
`2
`
`Smart Mobile Technologies LLC, Exhibit 2010
`Page 2 of 26
`
`

`

`to satisfy these requirements, but never gives more quality than required and necessary. Due to the
`dynamic wireless environment, adaptability of the system will be a key issue in achieving this. This
`implicates several mechanisms that can be used to attain a high energy efficiency, e.g.:
`•
`Avoid useless activity. This is the main driving force of for instance dynamic power management
`(Section 2.2), link layer protocols (Section 5) and adaptive error control (Section 6). Useless activity
`can be caused by various factors at all levels of the system (e.g. being unnecessary in a high power
`operational mode, applying error control to error-resilient data, trying to transmit a video frame that
`is already too old). If the operations would adapt to the required QoS and current environment, then
`energy-efficiency can be improved.
`Scheduled operations. This extents power management in the sense that communication is scheduled
`at appropriate time such that the differences in power states are exploited as much as possible
`(Section 4.2, and Section 5). This has a strong relation with the QoS model since timing constraints
`of multimedia connections are likely to be the limiting factors in the potential energy reduction.
`Reduce the amount of data. This is quite obvious and is applicable in all layers of the system. This
`relates to the trade-off between communication and computation (Section 3). Examples are adaptive
`error control that adapts its error coding according to the current channel conditions and the required
`quality, video transmission systems that adapt the quality to the expectations of the users, the
`available resources, and the channel conditions. Again, QoS can be used to determine whether it is
`really necessary to produce the data.
`
`•
`
`•
`
`Outline
`1.3
`In this paper we give an overview of various aspects of energy efficient wireless networking with a focus
`on the lower layers of network protocol stack. Although some research has been done in this area, a lot of
`research issues remain open. Because it is not possible to go into depth, the intention of this paper is
`primarily to give insight in the field of energy efficient networking.
`In Section 2 we provide the fundamentals of power management. Then, in Section 3 we review the main
`sources of energy consumption induced by the wireless channel. In Section 4 we provide an overview of
`mechanisms to reduce the energy consumption needed for communication in the network protocol stack,
`the operating system, and by decomposition. In Section 5 and Section 6 we will delve a bit deeper into the
`two most developed areas to reduce energy consumption: the MAC Layer, and error-control.
`
`2
`
`Power management
`
`Traditionally, energy efficiency has been focussed on low-power techniques for VLSI design. As the
`issue of energy efficiency becomes more pervasive, the battle to use the bare minimum of energy will be
`fought on multiple fronts: semiconductor technology, circuit design, design automation tools, system
`architecture, operating system, and application design. Energy awareness is now appearing in the
`mainstream digital design community affecting all aspects of the design process. Eventually, the concern
`for low-power design will expand from devices to modules to entire systems, including application
`software and even to the user.
`
`Low power system design
`2.1
`Most components are currently fabricated using CMOS technology. Main reason for this bias is that
`CMOS technology is cost efficient and inherently consuming less power than other technologies. The
`dominant factor of energy consumption (85 to 90%) in CMOS is dynamic. A first order approximation of
`the dynamic energy consumption of CMOS circuitry is given by the formu la:
`Pd = Ceff V2 f
`
`( 1 )
`
`where Pd is the power in Watts, Ceff is the effective switch capacitance, V is the supply voltage, and f is
`the frequency of operations. The power dissipation arises from the charging and discharging of the circuit
`node capacitance found on the output of every logic gate. Ceff combines two factors C, the capacitance
`being charged/discharged, and the activity weighting a, which is the probability that a transition occurs.
`
`Ceff = a C
`( 2 )
`
`
`
`3
`
`Smart Mobile Technologies LLC, Exhibit 2010
`Page 3 of 26
`
`

`

`
`
`At lower levels energy consumption can thus be decreased by reducing the supply voltage, reducing the
`capacitive load and by reducing the switching frequency.
`In general, a designer tries to make a system to be optimal for a certain application and environment. The
`designer has to select a particular algorithm, design or use an architecture that can be used for it, and
`determine various parameters such as supply voltage and clock frequency. However, energy efficiency in
`mobile systems is not only a one-time design problem that needs to be solved during the design phase. In
`a mobile system, power management extents the notion of hardware/software co-design, since we have to
`face a much more dynamic application and communication environment. When the system is operational,
`frequent adaptations to the system are required to obtain an energy efficient system that can fulfil the
`requirements imposed in terms of a general QoS model. This multi-dimensional design space offers a
`large range of possible trade-offs.
`
`Dynamic power management
`2.2
`The essential characteristic of energy consumption for static CMOS circuits is that quiescent portions of a
`system dissipate a minimal amount of energy. Dynamic power management refers to the general class of
`techniques that manage the performance and throughput of a system based on its computational needs
`within the energy constraint [6]. Dynamic power management exploits periods of idleness caused by
`system under-utilisation. Especially in mobile systems, the utilisation is not constant and power
`management can be used effectively. It is common practice that designers focus on worst-case conditions,
`peak performance requirements and peak utilisation, which, however, is in practice only fully exploited
`during a small fraction of their operation.
`Dynamic power management is based on deactivating functional units when they are not required. The
`main problems involved are the cost of shutting-down and restarting a module or component. Restarting
`induces an increase in latency (e.g. time to restore a saved CPU state, spin-up of a disk), and possibly also
`an increase in energy consumption (e.g. due to higher start-up current in disks). The two main questions
`involved are then: 1) when to shutdown, and 2) when to wake-up.
`1. The time (and thus energy) that is required to determine when a module can be shut down (the so-
`called inactivity threshold) can be assigned statically or dynamically. In a predictive power
`management strategy the threshold is adapted according to the past history of active and idle
`intervals.
`2. The other question is when to wake-up, where the typical policy is to wake up in response to a certain
`event such as user interaction or network activity. The problem with such a demand policy is that
`waking up takes time, and the extra latency is not always tolerable. Again, a predictive approach,
`where the system initiates a wakeup in advance of the predicted end of an idle interval, often works
`better.
`
`Operating modes of a wireless interface
`The wireless network interface of a mobile computer consumes a significant fraction of the total power
`[53]. Typically, the transceiver can be in five modes; in order of increasing energy consumption, these are
`off, sleep, idle, receive, and transmit (see Figure 1). In transmit mode, the device is transmitting data; in
`receive mode, the receiver is receiving data; in idle mode, it is doing neither, but the transceiver is still
`powered and ready to receive or transmit; in sleep mode, the transceiver circuitry is powered down,
`except sometimes for a small amount of circuitry listening for incoming transmissions [35]. The
`difference in the amount of energy consumed in these modes is significant.
`Examples:
`1) The power consumption of a WaveLAN modem when transmitting is typical 1675 mW, 1425
`mW when receiving, and 80 mW when in sleep mode [56]. Increasing the sleep time period of
`the radio thus significantly improves the energy efficiency of the wireless network. Also
`important to notice is that the transition times between the operating modes can be quite high. In
`WaveLAN a transition time from sleep to idle takes 250 ms, and has during that period already
`the power consumption of the idle state [21]. Then, before the payload will be transmitted
`another 254 ms is required.
`
`
`
`4
`
`Smart Mobile Technologies LLC, Exhibit 2010
`Page 4 of 26
`
`

`

`2) A Bluetooth radio (Ericsson PBA 313 01/2 [2]) has similar characteristics: in sleep mode it
`consumes 100 mA, in idle 25 mA, in receive mode 52 mA, and in transmit mode 44 mA. From
`sleep to idle requires 110 ms, and from idle to transmit or receive mode typical 104 ms.
`
`Energy efficiency
`2.3
`We define the energy efficiency e as the energy dissipation that is essentially needed to perform a certain
`function, divided by the actually used total energy dissipation.
`Essential energy dissipation for a certain function
`Actually used total energy dissipation
`
`e =
`
`
`
`( 3 )
`
`The function to be performed can be very broad: it can be a limited function like a multiply-add
`operation, but it can also be the complete functionality of a network protocol.
`Let us for example consider a medium access control (MAC) protocol that controls access to a wireless
`channel. The essential energy dissipation is the energy dissipation needed to transfer a certain amount of
`bits over the wireless channel, and the total actually used energy dissipation also includes the overhead
`involved in additional packet headers, error control, etc., but also 'physical' overhead induced by e.g. a
`frequency hopping scheme.
`Note that the energy efficiency of a certain function is independent of the actual implementation, and thus
`independent of the issue whether an implementation is low-power. Low-power is generally closely related
`to the hardware, whereas energy-efficiency relates to the algorithm using the hardware. Thus, it is
`possible to have two implementations of a certain function that are built with different building blocks, of
`which one that has been build with power-hungry components has a high energy efficiency, but dissipates
`more energy than the other implementation which has a lower energy efficiency, but is built with low-
`power components.
`
`3 Energy consumption in mobile systems
`
`Several researchers have studied the power consumption pattern of mobile computers. However, because
`they studied different platforms, their results are not always in agreement. Laptops designers use several
`techniques to reduce energy consumption, primarily by turning devices off after a period of no use, or by
`lowering the clock frequency. Lorch reported that the energy use of a typical laptop computer is
`dominated by the backlight of the display, the disk and the processor [36]. Ikeda et al. observed that the
`contribution of the CPU and memory to power consumption has been on the rise the last few years [27].
`Stemm et al. [53] concluded that the network interface consumes at least the same amount of energy as
`the rest of the system (i.e. a Newton PDA). Further, the fraction of energy consumed for networking by
`these mobiles, is only likely to increase as mobiles evolve towards a thin client network computer, and the
`communication traffic will increase. Another source of energy consumption is due to the fact that many
`high-performance network protocols require that all network access be through the operating system,
`which adds significant overhead to both the transmission path (typically a system call and data copy) and
`the receive path (typically an interrupt, a system call, and a data copy). This not only causes performance
`problems, but also incurs a significant energy consumption. Intelligent network interfaces can relieve this
`problem to some extend. To address the performance problem, several user-level communication
`architectures have been developed that remove the operating system from the critical communication
`path [7].
`To make the wireless interfaces more energy efficient, algorithms embodying energy efficient protocols
`must be distributed across two or more wireless end-points [32]. This implies that the focus should be on
`the layers of the network stack through which the mobiles interact.
`Even though it is difficult to compare these results because the measurements are made for different
`architectures, operating systems, communication interfaces, and benchmarks, there is a common pattern:
`there is no primary source of energy consumption, and the energy consumed for communication
`increases. The energy consumption is distributed over several devices and for several operations. The
`conclusion is that implementing an energy efficient system should involve all the functions in the system
`at all layers.
`
`
`
`5
`
`Smart Mobile Technologies LLC, Exhibit 2010
`Page 5 of 26
`
`

`

`
`
`•
`
`System architecture
`3.1
`In its most abstract form, a networked computer system has two sources of energy drain during wireless
`networking [32]:
`•
`Communication, due to energy spent by the wireless interface. Communication energy is, among
`others, dictated by the signal-to-noise ratio (SNR) requirements and the radio cell diameter.
`Computation, due to (signal) processing and other tasks required during communication.
`Computation energy is a function of the hardware and software used for tasks such as compression
`and forward error correction (FEC).
`For long distance wireless links (macro cellular), the transmit-communication energy component
`dominates. However, for short distance wireless links (pico cellular) and in harsh environments where
`much signal processing and protocol computation may be used, the computation component can be
`significant or dominant [57]. Broadly speaking, minimising energy consumption is a task that will require
`minimising the contributions of communication and computation, making the appropriate trade-offs
`between the two. For example, reducing the amount of transmitted data may be beneficial. On the other
`hand, the computation cost (e.g. to compress the data being sent) might be high, and in the extreme it
`might be such that it would be better to just send the raw data.
`As semiconductor technology improves, computation gets relative cheaper, whereas communication has
`much less advantage of the smaller feature size. Therefore, communication will get relatively more
`expensive. This property also holds for multimedia applications, even though these applications typically
`require a significant computational effort as well. For a significant part this is due to the limitations of
`most current hardware and operating systems that are unable to differentiate between various traffic
`streams [23].
`
`Adaptability
`3.2
`Recent research shows that by changing the system architecture from a traditional approach to a
`connection oriented, reconfigurable approach gives a huge improvement of the energy efficiency of a
`multimedia system [37][23][24][55]. Programmability is particularly important for mobile systems
`because they operate in a dynamically changing environment and must be able to adapt to the new
`environment. For example, a mobile computer will have to deal with unpredicted network outage or
`should be able to switch to a different network, without changing the application. It should therefore have
`the flexibility to handle a variety of multimedia services and standards (like different video decompression
`schemes and security mechanisms) and the adaptability to accommodate the nomadic environment,
`required level of security, and available resources. Reconfigurable computing systems combine
`programmable hardware with programmable processors to capitalise on the strengths of hardware and
`software. While low-power solutions are already available for application specific problems, applying
`these solutions in a reconfigurable environment is a substantially harder problem, since programmable
`devices often incur significant performance and energy-consumption penalties. To reduce the energy
`overhead in programmable architectures, the computational granularity should be matched to the
`architectural granularity.
`
`•
`
`Energy consumption of a wireless channel
`3.3
`We will now provide the main causes of unnecessary energy consumption needed for communication
`over a wireless channel ([22][51]).
`•
`First of all, for applications that have low traffic needs, the transceiver is idling most of the time.
`Measurements show that on typical applications like a web-browser or e-mail, the energy consumed
`while the interface is on and idle is more than the cost of actually receiving packets [53].
`Second, the typical inactivity threshold, which is the time before a transceiver will go in the off or
`standby state after a period of inactivity, causes the receiver to be in a too high energy consuming
`mode needlessly for a significant time.
`Third, in a typical wireless broadcast environment, the receiver has to be powered on at all times to
`be able to receive messages from the base station, resulting in significant energy consumption. The
`receiver subsystem typically receives all packets and forwards only the packets destined for this
`mobile. The access to the wireless channel is controlled by a MAC protocol. Many MAC protocols
`for wireless networks are basically adaptations of MAC protocols used in wired networks, and ignore
`
`•
`
`
`
`6
`
`Smart Mobile Technologies LLC, Exhibit 2010
`Page 6 of 26
`
`

`

`•
`
`•
`
`•
`
`•
`
`•
`
`energy issues [32]. For example, random access MAC protocols such as carrier sense multiple access
`with collision avoidance (CSMA/CA) and 802.11 typically require the receiver to be powered on
`continually and monitor the channel for traffic.
`Fourth, overhead induced by the physical layer. This overhead can be significant and is caused by for
`example guard space, interfacing delay, preamble and postamble.
`Examples:
`WaveLAN has an overhead equivalent to approx. 58.25 bytes [56] (see Figure 2). A frequency-
`hopping scheme makes this effect even worse, because it requires the radio to change its
`frequency often, like Bluetooth. In Bluetooth [1] a group of at most seven active slave radios, is
`synchronised to a single master radio. Slaves may only communicate when granted permission
`from the master. Bluetooth radios communicate with each other using a Time Division Duplex
`(TDD) scheme, whereby one radio starts transmission on even slots and the other on odd slots. A
`multislot packet can be extended over 3 and 5 slots (see Figure 3). Due to the frequency-hopping
`scheme and radio requirements the overhead to transmit data can be quite significant (i.e. the
`effective maximal use of the channel is for a 1-slot packet (30 bytes) 38%; a 3-slot packet (185
`bytes) 78%; and for a 5-slot packet (341 bytes) 87%).
`Fifth, another main contributor to overhead is due to the transition times between the various
`operating modes of the wireless radio. For example, the WaveLAN interface – that has a throughput
`of 2 Mbit/s –, already takes 250 ms (or virtually 62.5 bytes) to make a transition from sleep to idle.
`An obvious conclusion is thus that efficient data transmission (in terms of bandwidth utilisation and
`energy consumption) can only be achieved if the amount of data transmitted is not too small. A
`protocol that assigns the channel per slot will cause significant overhead due to turnaround, resulting
`in a significant energy waste. For example, in Bluetooth, transitions are required to be very frequent,
`and are thus causing a lot of overhead.
`Sixth, in broadcast networks collisions may occur (happens mainly at high load situations). This
`causes the data to become useless and the energy needed to transport that data to be lost.
`Seventh, the overhead of a protocol also influences the energy requirements due to the amount of
`'useless' control data and the required computation for protocol handling. Typical functions in the
`protocol stack include routing, congestion control, error control, resource reservation, scheduling,
`etc. The overhead can be caused by long headers (e.g. for addressing, mobility control, etc), by long
`trailers (e.g. for error detection and correction), and by the number of required control messages (e.g.
`acknowledgements).
`Finally, the high error rate that is typical for wireless links is another source of energy consumption.
`First, when the data is not correctly received the energy that was needed to transport and process that
`data is wasted. Secondly, energy is used for error control mechanisms. On the data link layer level
`error correction is generally used to reduce the impact of errors on the wireless link. The residual
`errors occur as burst errors covering a period of up to a few hundred milliseconds. To overcome these
`errors retransmission techniques or error correction techniques are used. Furthermore, energy is
`consumed for the calculation and transfer of redundant data packets and an error detection code (e.g.
`a CRC). Finally, because in wireless communication the error rate and the channel's signal-to-noise
`ratio (SNR) vary widely over time and space, a fixed-point error control mechanism that is designed
`to be able to correct errors that rarely occur, wastes energy and bandwidth. If the application is error-
`resilient, trying to withstand all possible errors wastes even more energy in needless error control.
`Another strategy is to reduce the data transmission rate or stop data transmission altogether when the
`channel is bad, i.e. when the probability of a dropped packet is high, so that less transmission time is
`wasted sending packets that will be dropped [59].
`
`4 Energy reduction in network protocols
`
`Energy reduction should be considered in the whole system of the mobile and through all layers of the
`protocol stack, including the application. Adaptability of the protocols is a key issue. The two basic
`principles to achieve an energy efficient system are: avoid unnecessary actions, and reduce the amount of
`data traffic.
`
`
`
`7
`
`Smart Mobile Technologies LLC, Exhibit 2010
`Page 7 of 26
`
`

`

`
`
`Network protocol stack
`4.1
`We will now provide an overview of how these basic principles can be used at the layers of a typical
`network protocol stack.
`•
`Physical layer – To allow a dynamic power management, we need to apply a radio that can be in
`various operating modes (like variable RF power and different sleep modes). Energy can also be
`saved if it is able to adapt its modulation techniques and basic error-correction schemes. The
`bandwidth offered by the radio also influences its energy consumption. The energy per bit
`transmitted or received tends to be lower at higher bit rates. A low bit-rate radio is less efficient in
`energy consumption for the same amount of data. However, when a mobile has to listen for a longer
`period for a broadcast or wake-up from the base station, then the high bit-rate radio consumes more
`energy than the low bit rate radio. Therefore, the low bit-rate radio should be used for the basic
`signalling only, and as little as possible for data transfer. This principle is for example used in
`HIPERLAN.
`Example:
`HIPERLAN [15] is the wireless LAN specified by the ETSI. Its energy saving is based on two
`mechanisms: a dual data rate radio, and buffering. Because HIPERLAN is based on a broadcast
`channel, each station needs to listen to all packets in its range. To decide whether the station is
`the destination of a packet, each pac