______________________________Fifth IEEE Workshop on Signal Processing Advances in Wireless Communication, Lisboa, Portugal, July 11-14, 2004 Energy-Scalability Enhancement of Wireless Local Area Network Transceivers Bruno Bougard 123 , Sofie Pollin 12 , Gregory Lenoir 1 , Wolfgang Eberle 12 , Liesbet Van der Perre 1 , Francky Catthoor 12 , Wim Dehaene 2 1 IMEC Kapeldreef 75, B-3001 Leuven Belgium 2 ESAT, K.U. Leuven Kasteelpark Arenberg 10, B-3001 Leuven Belgium E-mail: {bougardb, pollins, lenoir, eberle, vdperre, catthoor}@imec.be, wim.dehaene@esat.kuleuven.ac.be Abstract – Next generation Wireless Local Area Networks (WLANs) have to cope with energy budgets severely con- strained by portability, autonomy and high integration re- quirements. Practical power management approaches cur- rently implemented aim at reducing the transceiver duty cycle. However, recently developed energy-aware link adap- tation techniques, which trade off dynamically performance versus energy consumption, potentially bringing a factor-10 consumption reduction, promise to be more effective. Yet, to enable a meaningful trade-off, systems must present suffi- cient energy-scalability, i.e. energy consumption benefit when reducing the performance requirements or the envi- ronment constraints. This is not the case in current WLAN transceivers for which we show that duty cycle-based power management strategies are more effective. To make effective energy-aware link adaptation possible in future WLAN transceivers, we present techniques aiming at increasing their energy-scalability. Results show that a up to 7-fold energy consumption scalability can be achieved, providing significant margin to get energy consumption reduction by adapting to the user requirements. I. INTRODUCTION The recent success of Wireless Local Area Network (WLAN) solutions pushes developers to increase the data rate while supporting enhanced Quality of Services (QoS). On the other hand, the recent trend to have WLAN support in small form-factor, potentially multi-mode, bat- tery-powered, devices such as personal digital assistants (PDAs) makes energy-consciousness another strategic stake; therefore, the increasing importance of energy management techniques. In current WLAN standards (IEEE 802.11a/b/g), energy management is mainly based on duty cycle reduction using power saving states where the transceiver is partly deactivated [1]. A potentially more effective solution for energy management is to trade off energy consumption and performance using advanced energy-aware link adaptation techniques. In [2], it has been proven that exploiting dynamically (hence, at run- time) the energy-performance trade-off in so-called en- ergy-scalable systems (i.e. systems that finely adapt their energy consumption to the user requirements and to the environment constraints) may bring up to a factor-10 en- ergy reduction. In [3], the classical link adaptation prob- lem – i.e. exploiting optimally the channel capacity to maximize the throughput given an energy budget [4] – is reformulated to exploit the trade-off between link energy consumption and performance. Dynamic Modulation Scaling (DMS) and Dynamic Code Scaling (DCS) are 3 Bruno Bougard is granted by the Flemish Fund for Scientific Research (FWO) as Research Assistant introduced as practical way to achieve this goal. The pro- posed approach is extended in [5] considering jointly the adaptation to the channel profile and to the traffic statistic in order to minimize the transmitted energy. In [6], a more systematic approach is proposed to enable the de- sign of wireless systems that ensure a near-optimal adap- tive energy-performance trade-off, still with limited over- head. To make these techniques applicable to WLAN, the trans- ceivers have to present a sufficient energy-scalability, which is the energy consumption benefit when the per- formance requirements (e.g. data rate) or the environment constraints (e.g. propagation conditions) are relaxed. Yet, as shown in Section 2, this is hardly the case with current WLAN transceivers. In this paper, we propose techniques to make future WLAN transceivers energy-scalable. Next, we present a model to analyze the energy-performance trade-off at system level to finally study the impact of the proposed techniques on the global energy-performance trade-off. The remainder of this paper is organized as follows. In Section 2, standard WLAN systems are analyzed in order to find the parameters that have most impact on their per- formance and energy consumption. Then the modifica- tions needed to make them energy scalable are detailed. In Section 3, the modeling approach used to analyze the energy-performance scalability is presented (with a par- ticular attention to the impact of the channel) and the ex- ploration results are detailed. Finally, conclusions are drawn in Section 4. II. ENERGY-SCALABLE WLAN TRANSCEIVERS The energy-scalability of a system can be defined as the range in which the energy consumption can vary when the performance requirements - e.g. the user data rate - or the external constraints - e.g. the propagation conditions – vary from worst to best case. In this work, we consider the energy-scalability of OFDM-based WLAN such as proposed by the 802.11a/g standards. P Tx (8%) P PA (41%) P FE (20%) P DAC (6%) P DSP+MAC (25%) P FEC (35%) P ADC (15%) P DSP+MAC (25%) P FE (25%) Transmitter Receiver P Tx (8%) P PA (41%) P FE (20%) P DAC (6%) P DSP+MAC (25%) P FEC (35%) P ADC (15%) P DSP+MAC (25%) P FE (25%) P Tx (8%) P PA (41%) P FE (20%) P DAC (6%) P DSP+MAC (25%) P FEC (35%) P ADC (15%) P DSP+MAC (25%) P FE (25%) Transmitter Receiver Figure 1 Energy Consumption breakdown in typical wireless transceiv- ers. Obtained from the compilation of published energy consumption data for state-of-the-art 802.11a WLAN chipsets [7, 8]. It can be seen that the PA dominates the transmitter power while the receiver power is dominated by the digital signal processing, and more particularly the FEC decoding