Embedded controller considerations for WLAN apps

The general requirements of WLAN systems, to a large degree, are agnostic to the type of embedded controller that is used to enable them. Certain features of the embedded controller however, can be optimized for these types of systems.

In a typical WLAN configuration, a transceiver device, called an access point, connects to the wired network from a fixed location using standard cabling. At a minimum, the access point receives and transmits data between the WLAN and the wired network infrastructure. A single access point can support a small group of users and can function within a range up to several hundred feet. End users access the WLAN through special adapters, which are implemented as PC cards in notebook or desktop computers or integrated within PDAs and other devices.

System characteristics

WLAN applications have three dominant requirements. Firstly, and perhaps most importantly, cost. As these applications move from the corporate domain into the highly competitive SOHO market, the cost pressures on the system as a whole and the embedded controller element becomes more and more severe. Consequently, wireless gateway access point designs face significant cost pressures, placing severe constraints on the optimization of cost/performance tradeoffs and maintaining a high level of flexibility to accommodate system compatibility issues as air-interface and protocol standards are resolved.

Secondly, low power consumption. There are a few common approaches to reducing power consumption in wireless systems that use embedded controllers. Most suppliers of embedded controllers have squeezed as much of the power-hogging circuits as possible out of devices themselves. However, there still remains room to reduce overall system power budget. This can be done by integrating many of the functions that are usually implemented in standalone supporting chips, onto the embedded controller. This reduces power as the chip-count is reduced and there are less high speed switching circuits required in the overall system.

Recently, there have been several good examples of embedded controllers 'soaking up' the functionality of standalone devices or ASICs that usually surround it. A recent feature in embedded controllers is an integrated communications processor.

The third requirement for an embedded controller for WLAN applications is security. The embedded controller is designed to enhance system performance by executing computationally intense operations associated with the processing of IP security protocol (IPSEC), Internet key exchange (IKE), secure sockets layer (SSL) and wireless transport layer security (WTLS) protocols used in Broadband Access, customer premise equipment (CPE), routers, WAP gateways and other Access applications.

An example of this type of device is the MPC180 security processor which can be integrated into the system and will seamlessly operate with the "PowerQUICC". Thus, the host processor is not monopolized by the compute-intensive algorithms that are used for network security.

Embedded controllers

A typical embedded controller for a wireless system will include two key characteristics. These are an efficient pipeline to feed the CPU and a superscalar architecture. As it may not be obvious how these two attributes are useful for embedded controllers that are used in wireless systems, it is helpful to discuss the evolution of a typical embedded controller family. ColdFire emerged as a reduced instruction set computer (RISC) version of the popular 68K family. The 68K family was a popular device in wireless handsets and various computationally intensive control applications. The basic idea was to only implement the most commonly used instructions and addressing modes of the 68K on ColdFire but to implement them such that they were 'hardwired' and would, thus, run faster.

The subsequent enhancements in the ColdFire architecture were intended to reduce its size, increase its ability to feed the CPU and to execute more than one instruction at a time. The wireless segment heavily influenced these features.

ColdFire, being superscalar, allows more than one instruction to be executed concurrently. This ability is enormously useful for applications like wireless that require the execution of signal processing algorithms. In fact, the code that is executed in wireless applications is usually ideal for an enhanced pipeline superscalar architecture. Processing speeds of embedded controllers for WLAN systems will operate at around 300MHz today and will use additional 'advanced' architectural features such as a floating point unit (FPU) for complex arithmetic functions and an MMU to handle the interface of the memory sub-systems to the CPU.

Thus, while some CPU architectures have evolved to allow optimum performance for control-intensive code, the ColdFire architecture has evolved perfectly for operating the type of software that is used in wireless applications.

? Chris Platt

Ross Bannatyne