Achieving battery life for handhelds with 802.11b connectivity

Achieving battery life for handhelds with 802.11b connectivity
Handheld devices typically operate off of lithium ion or alkaline batteries, so power consumption is always a concern for system designers. Connecting to a WLAN is of little utility if the connectivity itself significantly reduces the battery life in the handheld unit.

Developers of handheld devices that use 802.11b connectivity find that they must consider design trade-offs in several areas. These include the device's operational range, data transfer rate, the type of power management techniques to employ and the probable usage cycle, which relates to the device's eventual end applications.

An effective way to understand the power requirements of a battery-operated system with 802.11b connectivity would be to construct a simulation of such a system. A Texas Instruments design team put together a model that demonstrates that a battery-operated product equipped with 802.11b connectivity can achieve reasonable battery life. It revealed that adding Wireless-Fidelity (Wi-Fi) capability does increase overall power consumption, but by an acceptable amount, and does not reduce battery life to the point where users would become unnecessarily inconvenienced.

When used in infrastructure mode, the user of a mobile 802.11b device can decide when to connect to a nearby WLAN access point (AP) and the conditions under which to maintain connectivity. There may be portions of the day when the user is out of range of a wireless network, or chooses not to be connected, such as when commuting or after business hours. These connectivity preferences can be stored in a device, such as a PDA, and used to optimize battery life.

In addition, the 802.11 specification itself gives mobile device designers considerable flexibility for implementing power management. While connected, the device can be placed in a power-saving "sleep" or standby mode between beacons, those instants in time when the 802.11b subsystem looks for information from the wireless network's AP. Beacons are transmitted at precise intervals, 10 times per second for example, and are used by the WLAN to identify all network members, and to alert these stations when data is waiting to be transmitted to them. This ability for Wi-Fi devices to enter a sleep mode allows a major reduction in average power consumption.

Taking this concept a step further, the device does not have to be awake for every beacon. The 802.11 protocol allows the station to use a parameter called the Listen Interval to save additional power. The Listen Interval is a parameter sent to the access point during network connectivity. The use of longer Listen Intervals allows the device to miss a specific number of beacons, without losing any data traffic or disconnecting from the network.

Under this mechanism, the AP buffers data while the station is asleep and not listening. You could se the Listen Interval at 10, for example, so that the 802.11b subsystem would wake to listen to every tenth beacon. Therefore, when 10 beacons occur per second, the device would wake and listen once every second. The device's response time using this Listen Interval would not be perceptible to users. The system would function as if it were continuously connected to the WLAN. It is important to note, however, that the AP and network must be able to handle data-buffering requirements for all associated devices.

Two important factors affecting power consumption in a mobile 802.11b system are the rate at which data is transmitted between the AP and stations, and the range or distance supported by the device's 802.11b radio frequency (RF). Range has a more straightforward relationship with power than data rate. A designer can specify a desired range for the device and reduce the RF transmit power if shorter distances are acceptable. In certain applications, it's possible to remove or bypass the RF system's power amplifier altogether. Doing so would reduce transmit power consumption, but it would also limit the supported data rate to 1 or 2 Mbits/second.

The data rate has a more complex relation to power consumption. For example, one approach to reduce 802.11b power would be to restrict operation to lower data rates (which are less susceptible to interference), in combination with lower RF transmit power. The same usable range, or distance, for the device could be maintained with reduced transmit power because lower data rates are more tolerant of interference.

Conversely, one may assume that higher data rates would result in higher power consumption, because higher transmit power is needed to overcome interference and maintain the same range than with lower data rates. However, we found that the duration of the transmission, not the power of the transmission, is the decisive factor. Our result curves are typical of Wi-Fi chip sets that are available in the market today.

With all things being equal, transmitting or receiving at speeds of 11 Mbits/s or greater requires less power than slower rates for a given amount of data. In addition, the 0dBm curves show that lowering the output power by removing or bypassing the power amplifier (and reducing the range) has a measurable, but not significant, effect on the power consumed. Thus, embedded designs will incur only a small penalty in battery life by using a full-power, full-range radio.

An examination of the power budget for a PDA with 802.11 reveals that the incremental power consumption and the resulting battery life are quite reasonable. The accuracy of any simulation depends upon the assumptions upon which it is built. In our test, activities involving the WLAN were defined for the PDA.

The activities defined were intended to be typical of a PDA with WLAN connectivity, such as e-mail, synchronizing the device with the user's desktop system, Web browsing and listening to voicemail messages. The model assumed that these activities would take place intermittently during an eight-hour workday, during which time the PDA would be continuously connected to the Wi-Fi network. For the remaining 16 hours of the day, the simulation assumed that the PDA would be in a low-power deep-sleep mode. Three user-activity levels with light, medium and heavy data loads were evaluated.

The model assumed that e-mail messages would be downloaded to the PDA 20 times throughout the course of a typical eight-hour business day. For every task the PDA performed, power consumption was calculated based on the total time that the device was performing the activity, the total number of transmit and receive bits that were transferred to or from the PDA during the course of each incident, and the average packet length of the data transmitted and received. The transmit and receive data transfer rates and the device's Listen Interval were also included. The model revealed that more than 80 percent of the time the Wi-Fi-enabled handheld was in a sleep or standby mode. The handhled spent the remaining 20 percent of the time in transmit or receive mode, which could be optimized with a higher data rate.

Assuming the use of a 3,600-mW-hour Li-ion battery (1,000 mA hours) and an average daily power consumption of 105 mW hours, we obtained a baseline battery life of about 34 days. With such results, adding Wi-Fi capability increases the power budget to approximately 140 mW hours per day. This would provide an expected battery life of more than 25 days, a net reduction in power consumption of 25 percent, which makes it very reasonable for enabling wireless LAN connectivity.