2002: time for more start-ups

Following a lull in the start-up frenzy of two years ago, one key entrepreneur believes Israel now needs more pioneers.

By David Greenfield

When 10-year old Doron Nevo babysat for Rami Oron, he could never have imagined it would lead to something bigger. Yet 20 years later Nevo, Oron and his father, Moshe Oron, are at Kilolambda, developing a DWDM system that will yield nearly 1000 wavelengths per fibre.

More wavelengths means, of course, that fewer fibres need to be laid - and that large-scale wavelength services in the metro can become a reality. In the future, technologies such as Kilolambda's could become key to unlocking the metro, says Stephanie Teral, director of European optical transport at telecoms consultancy RHK.

This kind of future-oriented thinking is common in the Israeli optical industry. Though set some 12,000km distant from Silicon Valley, Israel has developed its own version of the high-tech wonderland. In this country of 6.4 million people, there are some 50 optical start-ups.

Perhaps no company better exemplifies the highs and lows of the Israeli optical industry than Chromatis. Lucent's purchase of the metro optical manufacturer in 2000 for Euro 5.4bn (USD4.7bn) ignited the Israeli industry.

Yet the Israeli optical market has all of the key ingredients for long-term growth - research and development, funding, and experience. Nor are the current political problems posing a challenge for this technology hotbed. The situation might give buyers and investors some concern, but the impact on business is "negligible to non-existent" says Eyal Shekel, vice president and general manager of Chiaro Networks (Figs. 2 and 3).

Chiaro's solution has been to develop a massive optical switch (see Fig.1) with enough aggregate capacity to handle hundreds of line cards that may not be even be located in the same unit or building. The switch is an optical phased array device based on gallium arsenide that relies on the electro-optical effect to switch light. Chiaro plans to reveal exactly how it shifts the light channels around at OFC 2002. The company says it will be able to produce 8x8 or 16x16 switching elements that will enable up to 415x415 OC-192s to be switched in milliseconds.

With about 140 engineers per 10,000 workers, Israel has more per capita than anywhere else in the world. Many of those engineers and doctoral students are working in institutions such as the Technion, and universities of Ben Gurion and Tel Aviv.

University research has already led to a number of start-ups. Dan Sadot splits his time between teaching at Ben Gurion University, Beersheva, and working as the CTO of X-Light, a start-up based in Tel Aviv. X-Light is developing a subsystem that can rapidly tune multi-sectional lasers for use in a DWDM system. Multi-sectional lasers, like Agility's SG-DBR, provide a wide tuning range of more than 35nm. Single-section DBR lasers, by contrast, tune over just 10nm.

Like conventional lasers, multi-sectional lasers consist of an active cavity within which the lasing effect occurs. Instead of the typical reflective surface at either end of the cavity, however, these lasers incorporate gratings. The light built up in the cavity is passed through a grating which generates a specific frequency. By passing current across the grating, the grating's reflective index changes, altering the light's characteristics.

There are billions of possible combinations of currents that can generate a particular frequency. The light must be stabilised so that its frequency doesn't shift under ultra-fast dynamic conditions. Sadot has been able to arrive at a solution for tuning lasers to particular frequencies within 50ns.

Another professor-turned-entrepreneur, David Mendlovic of Tel Aviv University, claims that his company, Civcom (a start-up specialising in all-optical switching) was the first optical start-up to be spun out of the university. The company's Free-X line of optical switches provides switching using free-space propagation and polarisation encoding of optical signals. The result, says business development VP Jacob Vertman, is that the company enables ultra-fast, polarisation-independent, non-blocking switching. The company is introducing 1x2 and 2x2 switches with switching times of about 200ns.

Civcom's second core technology is what the company calls its Cobra Wave Processing technology, which combines a confidential mix of switching and filtering for a variety of applications.

Most new Israeli optical companies are looking to leverage their R&D skills to develop components or subsystems. More strenuous efforts to produce an entire system by the likes of Chiaro and Native Networks are relatively uncommon. "We systems providers are becoming an extinct species," says Native CTO Menahem Kaplan.

created one of the first WDM systems in 1997. The product used eight channels and was developed for the local phone company, Bezeq, to link Tel Aviv and Haifa. Bezeq reportedly never found the fibre to run the test, says Yossi Boker, COO of Red-C Optical Networking (a start-up specialising in tuneable optical amplifiers) and ECI didn't invest in the gear. Ciena came out with their system a year later and the rest is history.

ECI shook off that disappointment and moved forward. Through its Lightscape Networks subsidiary, the company sells DWDM and other optical networking gear around the world. Yet today entrepreneurs increasingly conclude that systems manufacturers don't necessarily play fully into Israel's perceived core competencies. Carriers are less willing to risk their networks on a start-up than they might have been a year ago - particularly one that might be long on R&D, but short on marketing and selling to a traditional telco.

Ron Shilon is CEO of Flexlight Networks, a supplier of passive optical network (PON) subsystems based in Petach Tikva. Shilon, who moved across from the IP voice world in September 2001, has a simple explanation of why Flexlight is going after the subsystem market. "It's just too difficult for remote, small start-ups to pitch the carriers directly," he says. Instead Flexlight's technology will be available to PON suppliers for integration into their own equipment or as a blade for integration directly into the metro gear.

Shilon says his PON solution is unique. Most PONS carry either ATM traffic, in the case of ATM PONs (APONs), or Ethernet traffic, in the case of EPONs, without requiring any active components between the switching centre and the subscriber. By avoiding having to power components in the fibre infrastructure, PONs can more than halve the cost of deploying fibre access.

However, there are problems with both approaches. The full cost reductions have not been realised with ATM. ATM gear is still quite expensive compared to Ethernet products. Performance is limited due to the segmentation and re-assembly needed when dividing packets into ATM's 53-byte cells. What's more, the ATM cell tax wastes bandwidth. The combination of these factors plus other limitations of ATM PONS, such as low speeds and lack of multi-cast support per service, have led to Ethernet PONs.

Yet Ethernet PONs carry their own limitations. Ethernet doesn't yet provide built-in quality of service (QoS) to ensure the delivery of toll-quality voice and high-resolution video. MPLS, DiffServ, and other mechanisms can be layered on top of Ethernet to provide these services, but only with higher cost and complexity. Also, Ethernet offers no operations, administration, maintenance, and provisioning (OAM&P) functionality, which limits the performance monitoring that a carrier can provide in the network.

Flexlight has instead developed what it calls the Multiprotocol PON (MPL-PON). The MPL-PON uses a proprietary algorithm that enables any two ends to a protocol-pipe to be connected together over a public network in their native formats. No complex conversions are required for circuit switched voice or Ethernet data. The MPL-PON, like an APON or EPON, is built along a synchronous network where a certain number of time slots are allocated and reallocated to each type of data using according to the defined service-level agreement. Through dynamic bandwidth allocation, time slots can be reallocated on the fly. OAM&P functionality is built into the protocol.

In the long term, prospects for the Israeli optical industry are as rosy as anywhere. A recent change in Israeli tax law should prevent foreign investors from being doubly taxed on investment in Israeli companies.

Zohar Zissapel, chairman of RAD Data Communications, has arguably produced more start-ups than any other entrepreneur in Israel. His biggest worry right now is that there aren't enough start-ups. "We had too many two to three years ago," he says. "But today few entrepreneurs are brave enough to open a company." He is concerned that in a few years there will be only middle-aged companies, but no "up-and-coming stars".

And if past record is any indication of the future, many of the companies will look to cash out early, a problem that Tzvi Marom, president and CEO of BATM (a manufacturer of optical switches, laments). "In Israel, entrepreneurs have become so infatuated with the fast exit that there is no opportunity to craft larger companies."

He believes Israel has become a sort of start-up-production line, generating high-tech start-ups that are subsequently acquired by the Nortels, Ciscos, and Lucents of the world. Then again, that may not be such a bad position after all.

-