The demands placed on customer premises equipment (CPE) in modern broadband networks are growing more complex by the day.

In Europe, fiber optic installations that reach all the way to the customer premises (“fiber to the home”) are becoming increasingly popular. At the same time, the data rates in both the networks of network providers (Wide Area Networks) and private networks (Local Area Networks, or LANs) continue to climb and, in the near future, will push broadband requirements to 100Mbps and higher. In addition, new applications – like TV and video on demand, as well as IP telephony – are contributing their share. For its part, the new Digital Living Network Alliance (DLNA) standard will enable a variety of digital content, such as photos, music, videos, etc., to be exchanged between popular home devices (TVs, hi-fi components, PCs, MP3 players, cell phones, game consoles, etc.), either via cables or wireless links. “Powerline,” which refers to the transfer of data within a LAN using the in-house power network, as well as data transmission via wireless LAN / WiFi (wireless Ethernet) are growing in importance because many homes aren’t equipped with the necessary telecom infrastructure, i.e., cables.

Today, the various CPE functionalities (router, wireless access point, voice-over-IP gateway, fiber-optic or TV-cable network termination unit, set-top box, etc.) are becoming increasingly integrated at the center of the home network. Some Internet service providers, in turn, are attempting to differentiate themselves from the competition, or to more closely bind customers, through their offering of Web services. Home automation, remote meter reading and access to digital content within the home via the Internet are just a few examples of such services. Indeed, service providers are currently developing systems that utilize Dynamic Service Deployment (DSD) to remotely deploy these types of applications in already installed CPEs – whenever the customer chooses, and without the trouble of opening the CPE device or installing software locally. In the meantime, there are also several dedicated organizations working on DSD solutions. The OSGi Alliance, for example, has specified a hardware-independent, dynamic software platform that facilitates the modularization and management (Service Registry) of applications and their services via a component model (Bundle/Service). Using the OSGi platform to realize such systems is much in favor today and requires a Java virtual machine upon which the OSGi framework can be built up. The Globus Alliance ( also offers implementations for DSD.

Those who have investigated these topics are quite aware that future-generation CPE devices will require a dramatic increase in both memory capacity and computing power. The system-on-chip solutions used in CPEs must therefore have enough resources to meet today’s known requirements, while at the same time holding enough computing power and memory in reserve for applications that will come along during a products’ lifetime.

Voice-over-IP telephony is still one of the “killer” applications for CPEs because it swallows up a great deal of resources. Here, too, standards have recently been upgraded – for example, wideband VoIP, a means of phoning via the Internet with even better voice quality. This is made possible by a sophisticated and complex voice coding method that efficiently, and to an extreme degree, compresses voice signals in accordance with the G.722 standard. However, it also requires more computing power. Another example of how technology is adding to the appetite for resources is CAT-iq, another new wideband standard for DECT-based telephones and IP-DECT base stations.

Broadband terminal devices in small to medium sized offices (home offices, one-person enterprises, small firms, etc.) require additional computing power for encrypted data transmission via virtual private networks (VPNs), while the security requirements of CPEs also mean that data packets must be inspected down to the last bit. CPE manufacturers and providers, for their part, prefer letting these applications run on a standard, low-power processor if possible – and not on difficult-to-program, power-hungry and expensive network processors.

To meet all of these requirements and also provide the needed computing power, MINDSPEED developed the Comcerto1000 system-on-chip solution, which it has since launched on the market. The Comcerto1000 is based on an architecture that utilizes dual ARM1136 CPUs, each with a clock rate of up to 650MHz. The new chip’s real “show stopper,” however, is the multi-layer bus, which connects both CPUs at extremely high speeds with all input and output interfaces. Data transfer from and to the input and output interfaces, and also between them, is performed via DMA, thus for the most part unburdening the CPUs.

The Comcerto1000 is equipped with all interfaces required for these applications, to include USB2.0, 2x PCI-Express, 2x Gigabit Ethernet and a DDR2 memory interface. As a result, the Comcerto1000 is able to serve as the heart of a modern hardware/software network platform. The PCIexpress (PCIe) interfaces support use of the newest, most compact wireless LAN modules i.a.w. the IEEE 802.11n standard, to enable data transmission rates of up to 248Mbits/s across distances of up to 250 meters in home networks. At the same time, PCIe supports the easy connection of serial ATA hard drives for memory applications, should the need arise.

The Comcerto1000 also offers quality of service capabilities – for example, ingress congestion, admission control, queuing and scheduling – which are realized as hardware in the MAC and thus further unburden the CPUs from computation-intensive tasks. Targeting CPEs in small and medium sized offices, the Comcerto1000 delivers the highest data packet throughput of any system-on-chip solution of its class, reaching 1Gbps in each direction even with the use of the encryption block for 3DES-SHA1 tunneling.

Typically, the (Linux) board support package installs the host operating system on the upper processor, or application & control processor (ACP), while tasks such as routing / bridging / NAT / firewall / QoS / multicast / IPSec and VoIP are assigned to modules on the media stream processor. The C1000 is capable of through-putting 2Gbps of aggregated data traffic – the maximum possible Gigabit-Ethernet speed on the line, in both directions – between the WAN and LAN with IPforward activated. At the same time, the application & control processor still has 95% of its performance available to handle the other services and applications mentioned above, with power consumption remaining below 2W!

MINDSPEED provides the Voice over IP software used with the Comcerto1000. The software allows for easy control of the system, thanks to the mature, tried and tested application programming interface (API) – which, by the way, is identical in design to the API used with the Comcerto 100 / 300 / 500 / 600 / 700 / 900 families. The VoIP software has been tested and optimized based on over 10 years of use in the media gateways of nearly all telecom manufacturers, and it offers numerous voice and wide-area codecs. Topping it all off is a G.168-2004 echo canceller that has also been tested a thousand times over during many years of use. Among others, it cancels echoes from multiple handsets, including echoes generated by cordless DECT handsets. Just one of the many advantages of MINDSPEED’s VoIP software is that customers do not need to worry about identification for most of the CODECS, since they are included in the MINDSPEED offering.

The Board Support Package is available for open source Linux OpenWRT, Jungo OpenRG, WindRiver Linux and TeamF1 SecureGW.

A typical CPE architecture built around the Comcerto1000 is shown in Figure 2.

To enable CPE manufacturers to use the Comcerto1000 in multiple applications while keeping the same system architecture in place, and thus to help save costs, the Comcerto1000 is offered in several pin-compatible versions. These differ in clock frequency and in whether or not they are equipped with a security accelerator block.

Users who choose the Comcerto1000 for their system can be sure that they are getting one of the most performing CPE platforms currently available in the world – one that has already proven itself in use for many customers around the globe.


Figure 1: Comcerto1000 Architecture
Figure 2: CPE architecture using the Comcerto1000
Hide this

This website uses cookies in order to give you the best possible service. By using this website you comply with the use of cookies. Further information can be found here